Scoreboard · every pick, graded
Every pick, graded.
Anyone can sound confident. We publish the exact kill line that would prove each pick wrong, then resolve it in public — played out, or invalidated. Strictly non-monetary; just whether the falsifiable claim was falsified. Below: every live commitment and every resolution, wins and losses alike.
Calibration · 350 scored
350 resolutions · directionalReliability
0.010
±7.4pp on 344 of 350
Picks resolved
351 resolved · 350 scored
160 reached · 191 closed on their line
Base rate
45%
picks reaching their case
Brier score
0.255
vs 0.248 for staking the base rate every time
The sample is 350 resolved picks. That clears the ~30-outcome line where a forecasting record stops being noise, and is nowhere near large enough to settle skill. An individual tier row with a handful of outcomes carries wide error bars, and discrimination between tiers needs a larger sample to measure precisely. The figures are published as computed and sharpen as the record extends.
What a Brier score of 0.26 means
A Brier score is the mean squared error between the probability a forecast stakes and what actually happened. Each conviction tier carries an implied probability of playing out — SUPREME around 90%, HIGH 75%, MEDIUM 60%, LOW 50% — and every resolved pick is graded against the odds its tier published in advance.
Read on its own the headline number is close to uninformative, because most of it is not a property of the forecaster at all. Under the Murphy decomposition a Brier score separates into reliability (0.010), resolution (0.003) and uncertainty (0.248). That last term is fixed by the question set: binary equity picks resolving at a 45% base rate carry an irreducible 0.248, which is exactly what a forecaster scores by staking that base rate on every pick and nothing more. It is the reference this record is read against — not a floor. A Brier score's floor is 0, and 0.255 sitting above 0.248 is what the negative skill score below states as a ratio. Comparing the raw figure against scores from geopolitical or general-knowledge forecasting compares base rates, not forecasters.
The two terms that do measure this desk are reliability and resolution. Reliability is a mean-squared term: the published 0.0104 is a weighted mean of SQUARED tier gaps, so its own percentage-point reading is its square root — 10.2pp RMS across every scored tier — and never 0.0104 read as a gap. That aggregate is dominated by whichever tier has fewest resolutions, which is why it is reported per tier instead. Across the conviction tiers carrying a meaningful sample (344 of 350 resolutions), stated confidence has tracked realised outcomes to within 7.4 percentage points. The remaining 6 sit in the HIGH tier, which is 58.3 points off its stated odds — too few outcomes to read as a rate, and the reason the aggregate reliability term is not a percentage-point gap. Every tier is listed below with its own count. Resolution at 0.003 is the open question: the tiers are honest about their own odds but do not yet pull outcomes far apart from the base rate, and separating those two properties takes more resolutions than 350.
The score is recomputable rather than asserted: the odds each tier stakes are fixed in advance, every resolved call is listed below with the tier it staked, and the arithmetic runs from the published record. One caveat stated plainly — a resolved pick is scored on the tier recorded when it was graded, and the research pipeline can restate a tier afterwards, which moves the figure with no new outcome behind it. Published resolutions are never restated to improve the score.
Does staking conviction beat the base rate yet?
Not measurably, at this sample size. The Brier skill score (−0.029) asks whether the conviction tiers carry information beyond simply forecasting the 45% base rate on every pick. It currently sits at the base-rate line, which is the same finding as the low resolution term above, expressed as a ratio. It is a statement about how far 350 resolutions can separate signal from noise, not a claim in either direction.
This is the distinction between a measured record and an asserted one. Research that retains its winners and quietly retires its losers can report any accuracy it chooses, because nothing was falsifiable at the outset. A figure computed over every resolved call — including the ones that hit their trigger — is one a reader can audit and recompute, which is what makes AI equity research checkable at all, and what an invalidation trigger exists to enable. Whether the record ultimately reflects skill or luck is a question this sample size cannot yet settle, and we publish the number that says so.
The Brier score, decomposed
0.010
Reliability: the gap between the play-out rate each tier claimed and the rate it actually hit. Lower is better.
0.003
Discrimination: how far the conviction tiers pull outcomes apart from the base rate. Higher is better; near zero means the tiers don't yet separate.
0.248
The irreducible difficulty of the question itself — the variance you'd face forecasting the base rate alone, before any conviction is staked.
These three sum to the headline figure: Brier = calibration − resolution + uncertainty. The split is the honest part. A low calibration term paired with a near-zero resolution term is exactly what a skill-≈0 record looks like — confidence that tracks reality reasonably well, but conviction tiers that don't yet sort the wins from the losses. The refinement term (0.245, uncertainty minus resolution) measures the same thing from the other side: how much sharper than the base rate the forecasts have managed to be so far.
Dashed = perfect calibration. Each dot is a conviction tier; the line shows the gap between what it claimed and what happened.
Does conviction predict the outcome?
| Tier | Claimed | Actual | n |
|---|---|---|---|
| HIGH | 75% | 17% | 6 |
| MEDIUM | 60% | 53% | 95 |
| LOW | 50% | 43% | 249 |
Play-out rate by archetype
- Compounder 53% · 36
- Cyclical recovery 31% · 70
- Theme leader 52% · 25
- Special situation 42% · 100
- Earnings inflection 56% · 64
- Retail squeeze 52% · 29
- Defensive 46% · 26
Play-out rate by regime at resolution
- Risk-on 46% · 328
- Neutral 41% · 22
351 resolved · 350 scored — one pick staked no conviction tier, so it cannot be scored for calibration
Two ways a pick resolves — played out or invalidated. Both stay on the record. A pick that has reached neither its target nor its kill line is undecided and is not scored: of 827 published picks, 160 reached their case, 191 hit their line, and 476 are still undecided. 42.4% of published picks have reached a verdict — 351 of 827. How a pick resolves →
- ADBE MEDIUM Played out
- AKAM MEDIUM Invalidated kill line fired
- ALAB MEDIUM Invalidated kill line fired
- AUR MEDIUM Invalidated kill line fired
- BHE MEDIUM Invalidated kill line fired
- DGII MEDIUM Invalidated kill line fired
- NET MEDIUM Invalidated kill line fired
- PLPC MEDIUM Invalidated kill line fired
- TTMI MEDIUM Invalidated kill line fired
- ULCC MEDIUM Invalidated kill line fired
- XNCR MEDIUM Played out
- AAON LOW Invalidated kill line fired
- ACDC LOW Invalidated kill line fired
- AIP LOW Invalidated kill line fired
- ALNT LOW Invalidated kill line fired
- AVR LOW Invalidated kill line fired
- BKNG LOW Played out
- BKSY LOW Invalidated kill line fired
- BMNR LOW Played out
- GFS LOW Invalidated kill line fired
- HTFL LOW Played out
- NP LOW Played out
- POWL LOW Invalidated kill line fired
- RH LOW Invalidated kill line fired
- SUPV LOW Played out
- TGTX LOW Played out
- EPAM LOW Played out
- TIC LOW Played out
- CYRX MEDIUM Played out
- IESC MEDIUM Invalidated kill line fired
- PCTY MEDIUM Played out
- WIX MEDIUM Played out
- GSM LOW Invalidated kill line fired
- HOV LOW Invalidated kill line fired
- LNC LOW Invalidated kill line fired
- SPSC MEDIUM Played out
- TEAM MEDIUM Played out
- G LOW Played out
- IQ LOW Invalidated kill line fired
- TASK LOW Played out
- CRWV MEDIUM Invalidated kill line fired
- ARM LOW Invalidated kill line fired
- INTC LOW Invalidated kill line fired
- CMCO MEDIUM Invalidated kill line fired
- FTK MEDIUM Invalidated kill line fired
- JWEL LOW Invalidated kill line fired
- KC LOW Invalidated kill line fired
- SAH LOW Invalidated kill line fired
- TLN LOW Invalidated kill line fired
- COHU MEDIUM Played out
- ENTG MEDIUM Played out
- EVER MEDIUM Invalidated kill line fired
- FA MEDIUM Invalidated kill line fired
- COCO LOW Played out
- CSCO LOW Invalidated kill line fired
- CVLG LOW Played out
- DUOT LOW Played out
- EXFY LOW Invalidated kill line fired
- FCEL LOW Played out
- HIMS LOW Invalidated kill line fired
- ICHR LOW Played out
- IHRT LOW Invalidated kill line fired
- MKSI LOW Played out
- MPC LOW Played out
- OEC LOW Invalidated kill line fired
- TSAT LOW Invalidated kill line fired
- VPG LOW Played out
- KEYS MEDIUM Played out
- PTGX MEDIUM Played out
- KLC LOW Invalidated kill line fired
- PRGS MEDIUM Played out
- WEX MEDIUM Played out
- RHI LOW Played out
- VLO MEDIUM Played out
- ASX LOW Played out
- BWIN LOW Played out
- CGC LOW Played out
- CRS LOW Invalidated kill line fired
- SATL LOW Played out
- CGEM MEDIUM Played out
- HTZ MEDIUM Played out
- SE MEDIUM Played out
- BBBY LOW Invalidated kill line fired
- BCAX LOW Invalidated kill line fired
- DJT LOW Invalidated kill line fired
- EVC LOW Invalidated kill line fired
- IMRX LOW Played out
- PANL LOW Invalidated kill line fired
- PBI LOW Invalidated kill line fired
- RUM LOW Played out
- CLF MEDIUM Played out
- ORKA MEDIUM Played out
- VG MEDIUM Played out
- EQNR LOW Played out
- FBIN LOW Invalidated kill line fired
- GLUE LOW Played out
- MGM LOW Invalidated kill line fired
- TORO LOW Played out
- WLFC LOW Invalidated kill line fired
- WSC LOW Invalidated kill line fired
- OKLO LOW Played out
- CCC MEDIUM Played out
- FRD LOW Played out
- SENS LOW Played out
- TNDM LOW Played out
- WEAV LOW Invalidated kill line fired
- WDC LOW Invalidated kill line fired
- ELF MEDIUM Played out
- MIRM MEDIUM Invalidated kill line fired
- RAPP MEDIUM Played out
- AKBA LOW Invalidated kill line fired
- BROS LOW Invalidated kill line fired
- CHYM LOW Played out
- FLY LOW Played out
- GGAL LOW Invalidated kill line fired
- GOLF LOW Invalidated kill line fired
- HELP LOW Played out
- LZ LOW Invalidated kill line fired
- NWPX LOW Invalidated kill line fired
- QURE LOW Played out
- ZVRA LOW Played out
- BBAR MEDIUM Invalidated kill line fired
- GHRS MEDIUM Played out
- COMP LOW Played out
- TBI LOW Played out
- UI LOW Played out
- HRI MEDIUM Played out
- NKTR MEDIUM Played out
- AEIS LOW Played out
- AOSL LOW Played out
- BRKR LOW Invalidated kill line fired
- DDOG LOW Played out
- ORIC LOW Played out
- VECO LOW Played out
- WGS LOW Played out
- AMBA MEDIUM Played out
- PGEN MEDIUM Played out
- RRGB MEDIUM Played out
- AAPL LOW Invalidated kill line fired
- AGX LOW Played out
- ATEN LOW Invalidated kill line fired
- BTE LOW Played out
- FIG LOW Played out
- KURA LOW Invalidated kill line fired
- MDB LOW Played out
- MSFT LOW Played out
- ODFL LOW Invalidated kill line fired
- RDW LOW Played out
- SABR LOW Played out
- SVV LOW Played out
- XRX LOW Played out
- VCYT MEDIUM Invalidated kill line fired
- NSIT LOW Played out
- RBLX LOW Invalidated kill line fired
- MRX HIGH Played out
- PRCH MEDIUM Played out
- KSS LOW Played out
- SNEX LOW Played out
- UNIT LOW Invalidated kill line fired
- IREN LOW Invalidated kill line fired
- NNE LOW Invalidated kill line fired
- CRDO MEDIUM Invalidated kill line fired
- CTOS MEDIUM Invalidated kill line fired
- MEI MEDIUM Invalidated kill line fired
- RMAX MEDIUM Invalidated kill line fired
- CAKE LOW Played out
- DDD LOW Invalidated kill line fired
- ENOV LOW Played out
- FOUR LOW Played out
- GTX LOW Invalidated kill line fired
- IART LOW Invalidated kill line fired
- JACK LOW Played out
- LUNR LOW Invalidated kill line fired
- ON LOW Invalidated kill line fired
- OUST LOW Invalidated kill line fired
- PAY LOW Played out
- SITM LOW Invalidated kill line fired
- RXO MEDIUM Invalidated kill line fired
- STX LOW Invalidated kill line fired
- UMC LOW Invalidated kill line fired
- MTUM LOW Invalidated kill line fired
- ACIW MEDIUM Played out
- SNDX MEDIUM Invalidated kill line fired
- ALHC LOW Invalidated kill line fired
- CDW LOW Played out
- FN LOW Invalidated kill line fired
- LILAK LOW Played out
- MTRN LOW Invalidated kill line fired
- TH LOW Invalidated kill line fired
- UCTT LOW Invalidated kill line fired
- VERU LOW Invalidated kill line fired
- EVH LOW Invalidated kill line fired
- LGIH LOW Played out
- M LOW Played out
- MYE LOW Played out
- MYGN LOW Invalidated kill line fired
- NEO LOW Invalidated kill line fired
- NOK LOW Invalidated kill line fired
- NUTX LOW Invalidated kill line fired
- REAX LOW Invalidated kill line fired
- SMCI LOW Played out
- VICR LOW Invalidated kill line fired
- AFRM MEDIUM Invalidated kill line fired
- MRCY MEDIUM Played out
- PANW HIGH Invalidated kill line fired
- BAND LOW Invalidated kill line fired
- GCMG LOW Invalidated kill line fired
- MMYT LOW Invalidated kill line fired
- XXI LOW Invalidated kill line fired
- ALGT HIGH Invalidated kill line fired
- FROG MEDIUM Invalidated kill line fired
- YETI LOW Played out
- AGYS LOW Invalidated kill line fired
- RARE LOW Invalidated kill line fired
- BULL MEDIUM Played out
- TWLO MEDIUM Invalidated kill line fired
- ZBIO LOW Played out
- BTGO LOW Played out
- CORT LOW Played out
- CPIX LOW Played out
- PLUG LOW Played out
- W LOW Invalidated kill line fired
- AMC LOW Played out
- AMKR LOW Invalidated kill line fired
- BB MEDIUM Invalidated kill line fired
- CLS MEDIUM Invalidated kill line fired
- ENPH LOW Invalidated kill line fired
- SANA LOW Invalidated kill line fired
- SPRY LOW Invalidated kill line fired
- FWRD LOW Played out
- EXTR LOW Invalidated kill line fired
- MARA MEDIUM Invalidated kill line fired
- PDFS LOW Invalidated kill line fired
- STM LOW Invalidated kill line fired
- TMC LOW Invalidated kill line fired
- KMX LOW Played out
- LEGN LOW Invalidated kill line fired
- BBY LOW Played out
- CLPT LOW Invalidated kill line fired
- NXT LOW Invalidated kill line fired
- PD LOW Played out
- JOBY LOW Invalidated kill line fired
- STUB LOW Invalidated kill line fired
- WING LOW Invalidated kill line fired
- AMD LOW Played out
- OXM LOW Played out
- PNRG LOW Played out
- NEXA LOW Played out
- PENG LOW Played out
- GCO LOW Invalidated kill line fired
- KALU LOW Invalidated kill line fired
- MTH LOW Invalidated kill line fired
- SHOO LOW Invalidated kill line fired
- FORM MEDIUM Invalidated kill line fired
- HIVE LOW Invalidated kill line fired
- KFRC LOW Played out
- KOS LOW Played out
- LGN LOW Invalidated kill line fired
- RMBS LOW Invalidated kill line fired
- VSH LOW Invalidated kill line fired
- MRVL MEDIUM Invalidated kill line fired
- IMOS MEDIUM Played out
- LPG LOW Played out
- BTDR LOW Invalidated kill line fired
- CIEN LOW Invalidated kill line fired
- FLEX LOW Invalidated kill line fired
- FPS LOW Invalidated kill line fired
- KEEL MEDIUM Invalidated kill line fired
- MTSI MEDIUM Invalidated kill line fired
- PI LOW Played out
- SAIL LOW Played out
- VIAV LOW Invalidated kill line fired
- ATI MEDIUM Invalidated kill line fired
- KYMR NONE Played out
- LRCX MEDIUM Invalidated kill line fired
- ONTO HIGH Invalidated kill line fired
- TNGX MEDIUM Invalidated kill line fired
- TSM HIGH Invalidated kill line fired
- MYRG HIGH Invalidated kill line fired
- AESI LOW Invalidated kill line fired
- AMN LOW Played out
- CCJ LOW Invalidated kill line fired
- HUT LOW Invalidated kill line fired
- ALTO LOW Played out
- SXC LOW Invalidated kill line fired
- SYRE LOW Invalidated kill line fired
- FAC LOW Played out
- AMSC LOW Invalidated kill line fired
- CLSK MEDIUM Invalidated kill line fired
- BTBT LOW Invalidated kill line fired
- SHLS LOW Played out
- PSNL MEDIUM Played out
- CEVA MEDIUM Invalidated kill line fired
- TLRY LOW Invalidated kill line fired
- RGTI LOW Invalidated kill line fired
- ADPT MEDIUM Played out
- GENI LOW Invalidated kill line fired
- NBR LOW Invalidated kill line fired
- QUIK LOW Invalidated kill line fired
- TSHA MEDIUM Played out
- NBIS LOW Invalidated kill line fired
- VRT MEDIUM Invalidated kill line fired
- OKTA MEDIUM Invalidated kill line fired
- SNOW LOW Invalidated kill line fired
- BFLY MEDIUM Played out
- DXCM MEDIUM Invalidated kill line fired
- MGTX MEDIUM Played out
- ZETA LOW Invalidated kill line fired
- CDNL LOW Played out
- BLZE MEDIUM Played out
- IPX LOW Invalidated kill line fired
- OPEN LOW Played out
- PTEN LOW Invalidated kill line fired
- HNGE LOW Played out
- RVMD MEDIUM Played out
- PRM MEDIUM Played out
- TBLA LOW Played out
- ALGM LOW Played out
- AUGO LOW Played out
- CECO MEDIUM Played out
- DIOD LOW Played out
- FIVN LOW Invalidated kill line fired
- JBLU MEDIUM Played out
- MCHP LOW Played out
- MRAM LOW Played out
- PCT LOW Invalidated kill line fired
- POWI MEDIUM Played out
- STRO LOW Played out
- TER MEDIUM Played out
- TRVI LOW Played out
- DRTS MEDIUM Played out
- TWST LOW Played out
- AZZ LOW Played out
- INDI LOW Invalidated kill line fired
- RIOT LOW Invalidated kill line fired
- CLFD LOW Invalidated kill line fired
- INOD LOW Invalidated kill line fired
- QCOM MEDIUM Invalidated kill line fired
- WEST LOW Invalidated kill line fired
- PURR LOW Invalidated kill line fired
- WTI LOW Invalidated kill line fired
- WULF MEDIUM Played out
- GLW LOW Invalidated kill line fired
- NVDA MEDIUM Played out
- RKLB MEDIUM Invalidated kill line fired
- MP MEDIUM Played out
- COHR MEDIUM Invalidated kill line fired
- GLXY LOW Invalidated kill line fired
- IONQ LOW Played out
- SGMT LOW Invalidated kill line fired
- APLD LOW Invalidated kill line fired
Open commitments · 42 · soonest to resolve first
Every pick on the active book — held, watchlist or rotation — that ships a published, falsifiable kill line. This is not the full scored set: a name we've stopped actively tracking keeps its published line on its own dossier and still resolves into the record above if that line fires, it just doesn't sit on this board. Distinct from the names held on the live book and the forward theme-bets scored on their own board.
-
Kill line A daily close below $868 hands back the entire August rebound off the 2026-08-11 close of $868.52 and returns price into the July range; a weekly close below $829 additionally breaks the 2026-08-04 gap-up shelf. Secondary: a formal US decision clearing Apple to source DRAM from CXMT.
-
Kill line A weekly close below $88.20 loses the shelf reclaimed on 2026-08-21 and re-opens the 2026-08-20 close of $85.31; secondary, 2026-09-21 passing with the $250M authorisation lapsed, $226.7M unused, and still no spectrum sale agreement announced since 2026-04-21.
-
Kill line A weekly close below $308 loses the 2026-08-10 post-print floor close of $308.82 and the 2026-06-18 breakout close of $314.44, leaving the 200-day printed at $250.45 on 2026-08-14 as the next reference; secondarily, the ~2026-08-26 short-interest report cutting the 2.31M-share base while price stays under the $358–360 zone.
-
Kill line A weekly close below $419 surrenders the July–August advance and re-enters the $405.00–$419.06 band where Silver Lake affiliates sold in July. Secondary: ISG operating margin printing under Q1 FY27's 10.5% at the 2026-09-01 Q2 FY27 report.
-
Kill line A weekly close below $238 breaks the 2026-08-13 shelf at $238.51, revisited on 2026-08-24. Secondary: the September Trump–Xi meeting passing with a published Apple allowance for CXMT DRAM or YMTC NAND and no company-dated rebuttal before the ~2026-11-04 Q3 report.
-
Kill line A weekly close below $234 forfeits the $235.67 shelf unbroken since 2026-07-08 and completes the round-trip of the advance off the 2026-06-25 print. Secondary: NVDA 2026-08-26, HPE 2026-09-02 and Dell 2026-09-03 all passing with no close back above $255.70.
-
Kill line A weekly close below $104 takes out the 2026-08-24 intraday low of $104.50 and leaves no part of the August advance intact. Secondary: the FCC September open-meeting tentative agenda (~2026-09-03) publishing with no transceiver-import item, which removes the policy option entirely.
-
Kill line A weekly close below $1740 surrenders the July band floor, confirms the failed reclaim of the $1,805 shelf from the 2026-08-14 settlement at $1,844.08 and re-exposes the $1,686.01 late-July low; secondary condition is the 2026-08-26 Nvidia print passing with no close back above $1,805 and no ASML-dated event until 2026-10-14.
-
Kill line A weekly close below $56 loses the early-August low of $55.74 and confirms the post-print leg is extending; secondary: no dated BlueBird 14/15/16 launch release by 2026-09-30, or beta service re-dated past 2026-12-31.
-
Kill line A weekly close below $62 completes the round trip of the 2026-07-31 post-Q2 gap and takes out the $64.25 April secondary price. Secondary: the 2026-08-26 NVIDIA print passing without a reclaim of the $73.8 shelf, or the Lumentum $43,500,000 initial deposit unconfirmed past ~2026-09-04.
-
Kill line A weekly close below $820 loses the 2026-08-11 pre-print shelf at $820.59, which held closing tests on 08-19, 08-20 and again on 08-24; secondary condition is the 2026-08-26 Nvidia call passing with no 2027 networking-content detail, leaving no priced event until the early-November FQ1 print.
-
Kill line A daily close below $171 breaks the floor of the $171.04–$179.01 August range that the 2026-08-24 close of $175.89 returned into and makes $179.94 a false high; secondary: Nvidia's 08-26 print, the ~08-25 FINRA report and Warsh's 08-28 keynote all passing with no close above the 2026-08-14 leg high of $180.18.
-
Kill line A weekly close below $8.22 returns price to the July 2026 $7.30–$8.22 range with the 50-day and 200-day both declining from their 2026-08-21 readings; secondarily an S-3 takedown, 424B5 or disclosed ATM usage dated after 2026-08-10.
-
Kill line A weekly close below $210 completes the round-trip beneath the $210.33–$211.27 July insider-sale range and puts price under the entire post-Q2 advance. Secondary: both the 2026-08-25/26 Jefferies and 2026-09-10 Benchmark-StoneX appearances passing with no reiteration of the $3.6B FY2028 framework.
-
Kill line A weekly close below $117 forfeits the post-gap floor set by the 2026-08-18 close of $117.18 and opens the $100 round number ahead of the 200-day at $95.27. Secondary: the 2026-08-27 Affirm print and the 2026-09-15/2026-09-17 conferences passing with $129.87 still unreclaimed.
-
Kill line A daily close below $17.84 ends the higher-low sequence off the 2026-07-29 low of $16.18 and reopens the pre-AT&T $16.21 shelf. Secondary: the 2026-08-27 Deutsche Bank appearance passing with no dollar value attached to the AT&T, Nasdaq Verafin or DOCOMO agreements.
-
Kill line A daily close below $8.64 (the 2026-08-18 range floor) re-opens the 2026-07-20 close of $7.96 and the 2026-07-17 low of $7.21; secondary: 2026-08-27 passing with the Clinch River comment record closed, no NuScale site named at TVA, and still no company release after 2026-08-05.
-
Kill line A daily close below $13.90 gives back the 2026-08-21 reclaim in full and re-exposes the 2026-08-06 shelf near $12.90, with the 2026-07-29 close of $10.74 beneath it.
-
Kill line A weekly close below $160 surrenders the 2026-08-19/20/21 policy gap (closes $160.20, $172.44, $186.49) and points back at the $139.11–$145.60 shelf; secondarily, the 2026-09-15 cloture motion on H.R. 3633 being pulled or finishing short of 60 votes.
-
Kill line A weekly close below $133 puts price back inside the 2026-08-07 to 2026-08-11 shelf and reduces the 2026-08-21 reclaim of the $135 IPO price to a one-session excursion. Secondary: the ~2026-09-09 tranche passing with another lower high and still no close above $146.15.
-
Kill line A weekly close below $104 surrenders the reclaimed July shelf and the entire post-summit advance, putting the 2026-08-14 close of $93.04 back in play; secondary confirmation is 2026-09-08 passing with STRC still under its $100 stated amount and no bitcoin purchases recorded.
-
Kill line A weekly close below $135.75 confirms the break of the twice-tested 08-11/08-20 shelf and leaves price under every disclosed August insider strike including Anderson's $134.87; secondary break if 2026-09-01 ISM and the ~2026-09-15 Cass August shipments pass without arresting July's -4.8% decline.
-
Kill line A weekly close below $80 surrenders the post-follow-on structure beneath the $85.00 June placement level, with no company disclosure due until ~2026-11-03; secondary conditions are August ISM Manufacturing under 52 on 2026-09-01 with New Orders below 50, or a downgrade or target cut out of the 2026-08-05 raise cohort.
-
Kill line A daily close below $50.24 takes out the early-August shelf and leaves no August structure under price. Secondary: a 2026-09-02 Q3 print with revenue under $11.5B, or FY26 non-GAAP EPS merely reaffirmed at $3.35–3.45 instead of raised.
-
Kill line A weekly close below $183 gives up the July–early-August shelf near $182.92 and the entire post-2026-08-10 upgrade advance; secondary is the 2026-09-02 Q1 FY27 print landing inside the $1.750–$1.900B revenue guide with FY27 reiterated at $7.325–$7.575B and no gross-margin raise.
-
Kill line A weekly close below $340 confirms the twice-failed reclaim of the $370 July shelf and the rising 200-day SMA marked at $369.20 on 2026-08-24. Secondary: Q3 FY26 on 2026-09-02 printing AI semiconductor revenue under the guided $16.0B or total revenue under $29.4B.
-
Kill line A weekly close below $186 breaks the rising 200-day that has floored the name all year, after the $197.01 shelf was lost on 2026-08-24; secondary condition is the ~2026-09-03 August LTL 8-K showing North American shipments per day decelerating under +5% or turning negative.
-
Kill line A weekly close below $21.53 hands back the 8/4 gap and the 8/3 pre-gap close, reopening the $19.16–$19.83 July shelf; secondarily, the 9/3 Q2 FY27 print cutting the $425–441M FY27 range, or Q2 revenue landing under the $102–107M guide.
-
Kill line A weekly close below $330 hands back the bulk of the August advance off the $297.38 52-week low with no 2026 base beneath it; secondary condition is 2026-09-03 passing with employee and invited riders only, no disclosed Cybercab fleet count and no date named for paid public rides.
-
Kill line A daily close below $372 takes out the 2026-08-11 close of $372.82 and turns the 2026-08-12 reclaim of the 50-day into a failed one; a weekly close below $353 removes the 2026-08-05 low of $353.69. Secondary: the 2026-09-08/10 conference window passing with no participation release and no intermediate 2027 figure.
-
Kill line A weekly close below $63 breaks the post-print sequence and confirms the failed retest of the 2026-08-14 close of $67.47; a weekly close below $56 would surrender the July range. Secondary: the September 8–10 healthcare-conference window passing with no incremental 2027 rate commentary while price stays under $67.47.
-
Kill line A weekly close below $259 converts the 2026-08-24 break of the $267.24 June floor from a policy gap into trend; secondary condition is the estimated 2026-09-09 DAT monthly release printing August dry van contract linehaul under $2.39/mile.
-
Kill line A weekly close below $167 takes out the 2026-07-30 low of $167.58 and the 200-day average marked at $166.56 on 2026-08-09 together, leaving the $119.32 52-week low as the next published shelf. Secondary: DAT's ~2026-09-09 August release showing van contract linehaul lower month over month after July's record +13 cents/mile.
-
Kill line A weekly close below $58.57 gives back the 2026-08-19 reclaim of the prior record close and returns price into the 2026-07-21 to 08-03 range; secondary: the ~2026-09-09 short-interest file showing the 13.22%-of-float base materially cut while the panel stays 10 Hold with nothing published above $65.
-
Kill line A weekly close below $96.43 confirms the loss of the 2026-07-21 shelf at $98.91, breached on the 2026-08-24 daily close, leaving $70.00 (2026-07-28) as the next dated closing reference. Secondary: 2026-09-10 Lake Street passing with no second distinct hyperscaler named for AI-ASIC wafer-level burn-in, after 2026-08-19 and 2026-08-25.
-
Kill line A weekly close below $267 hands back the entire post-print advance — the pre-print close was $267.25 on 2026-08-04 — and confirms 2026-08-14's $282.50 as a failed retest of the $281 shelf. Secondary: the ~2026-09-12 EO 62 working-group remit landing with no carve-out for co-located or behind-the-meter nuclear.
-
Kill line A weekly close below $97 forfeits the 2026-08-21 breakout above the $97.28/$98.87 band and reopens the $86.21 (2026-07-30) / $85.30 (2026-08-03) shelf. Secondary: the 2026-09-15 Senate cloture vote on H.R. 3633 failing or being pulled without a rescheduled date.
-
Kill line A weekly close below $78 unwinds the 2026-08-19 close of $78.59 that started the leg and gives back the 200-day reclaimed 2026-08-21, reopening the $70.80 50-day and the $66.67–$67.05 shelf; secondarily, CLARITY cloture failing 2026-09-15 with no SEC/CFTC substitute named, or Arc mainnet slipping past 2026-09-16 with no replacement date.
-
Kill line A weekly close below $30.11 gives back the 2026-08-05 print-eve close and returns price inside the 2026-08-06 gap-down; a weekly close below $26.54 retraces the entire post-print reaction. Secondary: the 2026-09-16 Investor Day passing with no 2027 membership range or premium framework.
-
Kill line A weekly close below $524 forfeits the 2026-07-30 low at $524.49 and re-exposes the $520.26 52-week low; secondary: Meta Connect on 2026-09-23 to 09-24 passing with no revenue-attached product or named external compute counterparty against the $130-145B capex guide.
-
Kill line A weekly close below $46.00 — 14.8% under the $54.00 contractual package and 5.7% above the ~$43.52 undisturbed pre-deal reference — marks break risk rather than collar mechanics; secondary conditions are sustained RKLB closes under the $67.50 collar floor, or the 2026-09-24 special meeting adjourning without an adoption vote.
-
Kill line A weekly close below $132.66 takes out the May low and the 52-week floor together, leaving no marked level beneath in this coverage window. Secondarily, a FERC order in ER26-3380-000 before the 2026-09-29 effective date confining accepted bids to new-build generation leaves the PJM backstop with no Vistra revenue attached.
How we're scored
A confident model is easy. A calibrated one is the point.
When a pick resolves, it lands in the ledger above as played out or invalidated, dated, with a flag for whether the published kill line actually fired. As outcomes accumulate, the calibration question gets a public answer: does conviction predict the outcome — does SUPREME beat LOW, does a regime call hold? Strictly non-monetary — only whether the falsifiable claim was falsified.
The odds each tier stakes · published in advance
- SUPREME 90%
- HIGH 75%
- MEDIUM 60%
- LOW 50%
Each tier claims this chance of playing out. The Brier score above measures the gap between these stated odds and reality.
Common questions
- Do AI stock predictions actually work?
- There's no honest one-word answer — only a record. FrontierPicks publishes one: every pick ships a dated, falsifiable kill line, and a real closing price is what settles it — played out, invalidated, or still undecided if neither line has printed. The ones that reached a verdict are scored with a Brier score. The point isn't a headline accuracy figure but that the calls are checkable: a model's predictions only "work" if they survive the lines they published, and that's exactly what the scoreboard above measures.
- How is FrontierPicks scored?
- Every pick ships with a published kill line. When it resolves it's marked played out or invalidated — dated, with whether the line fired. Across resolved picks we compute a Brier score (forecasting accuracy) and break play-out rates out by conviction tier, so you can see whether SUPREME actually beats LOW. Strictly non-monetary.
- How does a pick resolve, exactly?
- Mechanically, against the pick's own published kill line — no discretion. The reference is the real closing price the day the thesis was last stated. The kill is the price in the published line ("close below $X"); the target is a 1:1 realisation of that same published risk (as far above the entry as the kill is below it). Walking real daily closes forward, whichever is hit first decides it: a close at/below the kill is invalidated, a close at/above the target is played out. "Weekly close" lines are graded on weekly closes. A pick whose stop is within 2% (too tight to grade on price) or whose kill isn't a clean price level stays open. Graded only on whether the published claim was falsified.
- Why does the score start in June 2026?
- We rebuilt the research pipeline in June 2026, and a pick is graded on the thesis as the rebuilt pipeline stated it — from the day that thesis was written, not from the day the ticker entered coverage. Those are different dates for most of this record: 210 of the 351 resolved picks were first published before June 2026, and 14 carry an outcome date from before it, backfilled from outcomes the previous system had already written. So this is not a clean slate — it is a record of theses restated by the current pipeline, over a corpus much of which it inherited.
- What is a Brier score?
- The mean squared error between the probability a conviction tier implies (SUPREME≈0.9 … LOW≈0.5) and the binary outcome. 0 is perfect, 0.25 is a coin flip. For a rough reference band: expert human forecasters in Tetlock's Good Judgment work are commonly cited around 0.08, and the strongest LLM forecasters in 2024–25 evaluations around 0.10. Those benchmarks score different questions, so treat them as scale markers, not a like-for-like comparison.
- What does the scoreboard measure?
- Whether each falsifiable claim was falsified — graded against its own published kill line. It scores the reasoning, not a trade.
- What happens when a pick is wrong?
- It's marked invalidated, dated, and kept on the public record — with whether the published kill line fired first. Being wrong in public, on the record, is the point: it's what makes the score trustworthy.