Frozen historical comparison · do not promote
Historical comparison · negative result
Operator-reported retrospective cost sensitivity. Independent population validation pending.
do_not_promote_this_candidate
- Reported evaluation dates
- 2026-01-01 to 2026-06-13
- Games
- 813
- Admitted quotes (reported population)
- 812
- NBA teams represented
- 30
- Candidates
- 405
- Passes
- 407
- Excluded games
- 1
- Elo model Brier
- 0.205283084
- Identical-contract market Brier
- 0.185892087
Identical-contract Kalshi candle-close bid/ask midpoint reference, not an executable midpoint fill; lower Brier is better.
Not verified historical exchange returns, forward performance or live signals. No orders are placed. Real-money trading is not authorized.
Fixed Elo is less accurate than the same-contract market reference and all predeclared cost scenarios lose money in the evaluated population.
Population and exclusions
Development: 2025-04-19 to 2025-12-31; evaluation: 2026-01-01 to 2026-06-13. These are reported retrospective partitions, not five full seasons.
Teams: ATL, BKN, BOS, CHA, CHI, CLE, DAL, DEN, DET, GSW, HOU, IND, LAC, LAL, MEM, MIA, MIL, MIN, NOP, NYK, OKC, ORL, PHI, PHX, POR, SAC, SAS, TOR, UTA, WAS.
- stale_candle: 1
Exclusions here are within the reported evaluation population, not a full ingestion waterfall. Upstream unmatched contracts and missing periods have not been independently reconciled by this display.
Model minus market Brier: 0.019390996; reported game-bootstrap 95% interval: 0.011370 to 0.027989.
Hypothetical negative cost scenarios
12 labels / 8 distinct numerical cases for the fixed one-contract rule. All point estimates are negative; this does not prove every expected return is negative. Some reported intervals include zero.
| Scenario / adverse price allowance | Hypothetical net $ | Hypothetical cost $ | Hypothetical ROI | Hypothetical fills / no-fills | Reported net per candidate 95% interval ($) |
|---|---|---|---|---|---|
| quadratic_order_cent|slippage=0 | -15.79 | 158.79 | -9.94% | 405 / 0 | -0.083728 to 0.004556 |
| quadratic_order_cent|slippage=0.01 | -19.87 | 162.87 | -12.20% | 405 / 0 | -0.093731 to -0.005447 |
| quadratic_order_cent|slippage=0.02 | -23.95 | 166.95 | -14.35% | 405 / 0 | -0.103707 to -0.015522 |
| quadratic_order_cent|slippage=0.05 | -36.33 | 179.33 | -20.26% | 405 / 0 | -0.134248 to -0.046118 |
| quadratic_contract_cent_reserve|slippage=0 | -15.79 | 158.79 | -9.94% | 405 / 0 | -0.083728 to 0.004556 |
| quadratic_contract_cent_reserve|slippage=0.01 | -19.87 | 162.87 | -12.20% | 405 / 0 | -0.093731 to -0.005447 |
| quadratic_contract_cent_reserve|slippage=0.02 | -23.95 | 166.95 | -14.35% | 405 / 0 | -0.103707 to -0.015522 |
| quadratic_contract_cent_reserve|slippage=0.05 | -36.33 | 179.33 | -20.26% | 405 / 0 | -0.134248 to -0.046118 |
| double_coefficient_contract_cent_stress|slippage=0 | -22.02 | 165.02 | -13.34% | 405 / 0 | -0.099286 to -0.010641 |
| double_coefficient_contract_cent_stress|slippage=0.01 | -26.18 | 169.18 | -15.47% | 405 / 0 | -0.109485 to -0.020886 |
| double_coefficient_contract_cent_stress|slippage=0.02 | -30.32 | 173.32 | -17.49% | 405 / 0 | -0.119660 to -0.030983 |
| double_coefficient_contract_cent_stress|slippage=0.05 | -42.75 | 185.75 | -23.01% | 405 / 0 | -0.150348 to -0.061794 |
Uncertainty and remaining verification
- Historical fee applicability: unverified. Fee rounding/coefficient choices are scenarios, not actual charged fees.
- Actual tipoff: unverified. Later-retrieved scheduled starts are not actual tipoffs.
- Contemporaneous availability: unverified. Revised historical model inputs do not establish pregame publication.
- Executable depth, queue position, size and actual fills: unknown. Hypothetical fills are not executions.
- Bootstrap intervals are operator-reported sampling uncertainty, not independent validation or evidence of live capacity.
- Examined-period warning: this evaluation is now observed, not an untouched holdout. No threshold retuning claims on this same evaluation; future confirmation requires a genuinely separate period.
- Not verified historical exchange returns, forward performance or live signals. No orders are placed. Real-money trading is not authorized.
- Retrospectively selected partition, not a sealed prospective test; no threshold tuning performed.
- Quotes use scheduled start from later-retrieved ESPN records, not verified actual tipoff.
- Historical fee applicability and executable depth/fills remain unverified; returns are scenarios only.
- Model uses revised historical NBA results, whose original publication times are unknown.
- A fixed one-contract-per-game rule does not estimate live capacity or profit.
Frozen rules and provenance
- Original complete summary SHA-256
- 21c216f10b843f6ce2e43da504cb3ea7ebb9ca96c163ab997892bfaec5fbed56
- Original complete decision SHA-256
- a4ae16deda0704b9511c1a0da90671f0b16a709a668eadd6c533ce4dcac63425
- Summary created at
- 2026-10-04T22:37:37.375856+00:00
- Decision created at
- 2026-10-04T22:39:00.744002+00:00
- Decision's separate report SHA-256 (not the summary hash)
- fab2147cbe9b862dfdf7ba5f1b4b697f5e808c6fd69699dbeea23369bf7d8829
- Frozen comparison version
- kalshi-pregame-retrospective-v1
- Source classification
- research_cost_sensitivity_not_verified_executable_returns
Decision report hash is retained as reported; its separate report bytes are not staged here. The two complete staged inputs are pinned independently, not relabelled as a matching hash pair.
Retained frozen comparison rules (not new strategy rules)
- contract selection
- Lexicographically first finalized matched ticker per game, before examining quotes or outcomes; either YES or NO may be chosen, at most one contract per game
- decision minutes before scheduled start
- 30
- max candle age seconds
- 120
- model
- Existing fixed daily-batch Elo: base1500 K20 home100 scale400 annual reversion0.25; no tuning, games strictly before game calendar date
- selection
- Choose largest model probability minus ask minus 0.07 quadratic per-contract cent-ceiling fee reserve minus $0.01 adverse price allowance; require surplus>=0.05. Side ties choose YES. No rescreening under cost sensitivity.
- settlement
- Kalshi settlement_value_dollars (NO=1-YES), not NBA outcome substitution; no settlement-fee claim
- execution
- Hypothetical purchase at completed-candle close ask plus adverse sensitivity; no proof of available size or fill. Out-of-range price>=1 is no-fill; never use candle high/low as an achievable fill
- uncertainty
- Unique-game bootstrap with fixed seed73 and 1000 replicates; no complementary-contract double counting