What a Prop Firm Eval Pays If You Have No Edge

D18 · Test locked 2026-10-10 · Dated rules, conditional results

1. Question, scope and conditional registered answer

With no edge, through the first payout, on BluSky Launch 50K, rules retrieved 2026-10-10, the QQQ stand-in returned 58.2–64.2 cents per fee dollar at one trade a day. The simple model returned 59.8 cents. These are ratios of mean payouts to mean fees over purchases, including failures.

The price includes $59 per 30 days during evaluation, $99 on a pass, no paid reset, and 90% of a first payout of at least $500. Later payouts are a separate report-only result below. No edge means random entries, not a judgement about a trader.

We did not test JJ Simon's entries, skill or payouts. QQQ stands in for NQ futures; it is an assumption, not a verified futures replication.

2. JJ Simon's frame, source credits and what we grant

The claim comes from JJ Simon on Chart Fanatics, published 2026-10-04. He reports a 33% pass rate and a 91% chance of no payout; the complement is 9%. Those are his figures, not verified personal performance.

We grant his frame: a pass is a different objective from trading well; multiply pass probability by the chance of payment after passing; value a funded account over its paying life. His statement about value after payment ends here: “Because you still have expected value in your account after taking a payout. So net, you’re making a bunch of money”.

The source video’s title card says “$1.8 MILLION”. That is an attributed claim; this test does not audit the earnings. He names BluSky for his funded accounts, not the Launch plan for his described evaluation.

3. Locked plan, priors, decision rules and E3's failure

The main terms of the test plan we locked first are available as a free design summary, with the grid, decision rules, seeds, assumptions, checks and post-lock amendments. This is a summary, not the plan text. Lock 60d2486 was recorded before simulation on 2026-10-10 in a private repository; no public ref pins it. The original locked file hash and the later amended checkout hash are distinguished in the download. Later corrections and reading decisions are labelled beside the affected terms.

E3 failed: the edge change moved the pass probability more than the risk-size change. The risk comparisons describe model structure; they are not sizing advice.

4. Data, two arms, two cadences and dated firm's rules

Post-lock addendum 4 corrected the calendar after headline results. Accounts starting 2011-03-23 to 2023-06-14 use the QQQ/SPY common calendar. At lock, starts began at 2004-01-02 with a 756-session runway before 2026-06-22. Earlier QQQ sessions are separated by a data hole; the original early-era comparison is empty. QQQ and SPY are correlated stand-ins, and neither tests NQ directly.

The model arm has 200000 simulated purchases per cell; the path arm has 10000. These are uncertainty denominators, not advice on buying evaluations. C1 is one trade a day; C2 repeats entries until three consecutive losses, a phase target, a breach, or the path window’s end. The model has no intraday clock.

In the headline setup, evaluation and Buffer use 25/38-point brackets; Sim Funded uses 25/100. One NQ-equivalent risks $500 before $5.74 round-trip costs. Random entries use New York’s first 90 minutes. Path fractions use an estimated reference level of 30,000, checked at 24,000 and 36,000. The minimum contract limit below a $1,500 funded balance is unknown; the model retains one contract.

Intervals are 95%: Wilson proportions and bootstrap ratios/differences for the model; start-month block bootstrap for paths. Bars touching both levels are resolved stop-first, target-first and 50/50. The free design includes fills, gaps, caps and missing-bar rules.

5. Registered Q1-Q4 results, intervals, labels and flags

With no edge, through the first payout, on BluSky Launch 50K, rules retrieved 2026-10-10. C1 means one trade a day; C2 means many trades a session. Arm A is the simple model; B is QQQ at the primary reference level. The s/t/c suffixes mean stop-first/target-first/50/50. Q1 is BELOW both comparison bands in these eight cells; Q4 is OVERPRICED under each cadence and arm.

The quick sum omits the Buffer and funded failures. Film rounding “about $86” uses the stop-first fee. The paid-only phrase “about $1,440–1,450” is film rounding; the table shows the exact range, including $1,434.57. A paid purchase has incurred at least $158 in fees ($59 + $99).

The film’s “only one” above a dollar refers to no-edge returns under trailing-floor rules: 1 of 131 cells in a set defined after the results to check this sentence, selected by their parameters, not their outcomes. This is not a registered set. The above-dollar cell is the report-only C1 lifetime variant. Static-floor counterfactuals and edge-grid cells are outside that set; the quick sum is a different formula, not a measured return. Defined after the results to check the film's "only one" sentence; not a registered set: 131 no-edge, trailing-floor cells chosen by parameters. The one whole-period point above a dollar is B_multipay_C1, a report-only variant. Whole-period cells and point estimates only. Registered/check cells and report-only variants with no edge and a trailing floor are included; edge-grid cells, static-floor counterfactuals and the quick-sum formula are excluded. The era splits in the design download are sub-period readings and are not members. Among those excluded splits, B_discount_C1 for 2011-2017 starts is a report-only half-price variant: with no edge, on BluSky Launch 50K dated 2026-10-10, QQQ stand-in stop-first, one trade a day, through the first payout, its EVF point is 1.03 [0.91, 1.16] (95% interval), TOO CLOSE TO CALL. Its point exceeds a dollar, but its interval straddles a dollar. This split is not a second member of the whole-period set.

Q2 is model-only, through the first payout, on the dated plan. Its random-entry baseline is 39.68%. The threshold requires the lower bound above one at that grid value and every larger value. The alternative keeps 1.52 in the funded phase.

Q3, with no edge on the dated plan, has no single verdict: ARMS-DISAGREE in both cadences; the QQQ answer is CADENCE-SENSITIVE. D1 compares 1-to-1.52 with 1-to-1; D4 compares it with 1-to-4. TE subtracts the static-floor difference. Numbers below are percentage points with intervals.

In model C1, the trailing floor accounts for most of the gain against 1-to-1, but the static floor favours his bracket more against 1-to-4; hence NOT THE TRAIL overall. In model C2 the gain is similar under a fixed floor. QQQ C1 is FALSE against the 1-to-4 leg at the three reference levels and in 2011–2017; 2018–2023 alone is inconclusive (ERA-SENSITIVE). QQQ C2 varies by same-bar resolution (BAR-COARSE).

6. Every result cell and every spoken/on-screen figure with provenance

Every result of the corrected runs is traceable below. The defective first runs and their progress values are excluded and never quoted here. Superseded addendum-1 unit-test attempts and extra sanity re-runs are summarised as corrections, not reprinted. The final reported checks are included. This is the free report linked in the description. The ledger contains 294 distinct result cells; the film table contains 191 numerical sentence or screen-placement records. Animation repeats are coalesced; spoken and shown placements are distinct uses of evidence, not additional experiments.

The CSV contains full-precision values, denominators, parameters and conditions. A grid cell with an edge probability is an edge test, not a no-edge result. Static floors are counterfactuals. Per-cell EVF classifications are diagnostic and cannot substitute for the combined Q3/Q4 decisions.

Download the result ledger

The source-frame crop in B05-s09, from JJ Simon on Chart Fanatics at 8:28 (508.4 seconds), shows his whiteboard: his figures, shown as he wrote them; not our results. It shows $750. Our B05-s03 caption says +$760, following his spoken "-500 plus 760" at 1:06:08 (3968.96 seconds), with a 38-point target on one NQ. These identify the written and spoken sources; we do not decide here which he intended. The tiny red annotation above -50 on the whiteboard is not legible at 1080p.

The attributed title crop shows $1.8 MILLION. Its tiny background dashboard identifiers, dates and amounts are not legible at 1080p and are not transcribed as numerical claims.

Download the film figures

Words such as “first”, a date, a list count or a pronoun are retained when they occur in a numerical sentence. They are not automatically classified as measured results. The reset count, reset fee and fee-dollar units remain in the spoken rows even where their numeral was removed from the picture.

7. Report-only variants and modelling-choice directions

Report-only: every payout to the review. With no edge, on the dated plan, QQQ stop-first, the multi-payout measure averages across purchases until a breach or the $10,000 review, using the primary funded-floor reading. C1 is one/day; C2 is many/session. The latter point is below half the fee but its interval crosses half. Neither is the registered first-payout answer.

With no edge, through the first payout, on BluSky Launch 50K dated 2026-10-10, QQQ stand-in stop-first paths and the primary funded floor, the discount variant uses the published 50% first-account subscription price, $29.50 per cycle while it runs; C1 (one trade a day) is too close to call. Under that same first-payout scope, the Buffer-reset variant includes up to three $250 resets in fees. C2 repeats trades within the session under the stated stop rules. These change the priced conditions and are not purchase recommendations.

Of the three choices we measured against the no-edge QQQ stop-first, one-trade-a-day headline on the dated plan, first-payout scope has the largest effect on payout per fee dollar. The funded-breach choice was not measured; zero censoring is neutral by construction. The result ledger includes 37 path report-only cells, plus the model variants. The design download preserves parameters for the 95 report-only cells (58 model, 37 path), with the recorded checks. Era splits cover the 37 path report-only cells; the model has no era splits. Paired differences cover the registered Q3 comparisons (12 model and 432 path), not the report-only variants.

8. Limitations and four corrected mistakes, with amendment history

The test prices one dated plan and its modelled rules. It does not establish a personal edge or future returns. Minute bars hide touch order; futures basis, overnight trading and liquidation slippage are not modelled. The floor and consistency pages admit rival readings. SPY is correlated, and the path arm cannot validate its proxy against futures data we do not hold.

Four corrected mistakes correspond to amendments 1, 2, 4 and 5. Amendment 3 records choices made after two nonheadline progress values were visible. The defective first model/path runs are excluded from this report. The fifth amendment changed reported labels, not simulation outputs. The Q3 label printed here uses that corrected record.

9. Five reopen conditions, source/download links

Source record hashes identify private research records in their recorded UTF-8 bytes; they are provenance receipts, not public copies of those records. The design summary names the LF lock blob separately from the CRLF amended checkout.

Download the source registry Download the design summary and diagnostics

10. Educational notice, funding firewall and next test

Ground Truth is educational and entertainment content, not financial, investment or tax advice. Backtests and simulations are hypothetical, model costs imperfectly, and do not guarantee future results. Nothing here recommends buying evaluations or any security or strategy.

We never take money from anyone whose claims we test. Support links are separate from research. The next test asks whether a published entry has an edge: the opening range breakout, a first-hour setup.