← Nashpoint39 trades

The poker↔trading bridge

A poker hand review asks: was the decision right, independent of how the cards fell? This tab asks the same of a trade. First it scores every trade against a fixed decision-error taxonomy; then it runs AIVAT — the same variance-reduction estimator Nashpoint uses to grade poker — treating the market path as the chance node, so luck is stripped and true edge shows up in a small sample.

Luck-corrected edge · 34/39 trades graded

Raw P&L / trade

−$102.30

± 28.88 SE · 95% CI [−$158.90, −$45.70]

std ±168/trade — path noise drowns the signal

AIVAT edge / trade (luck-adjusted)

−$74.69

± 11.56 SE · 95% CI [−$97.35, −$52.03]

6.2× variance reduction → statistically negative edge

Raw per-trade P&L can't tell bad luck from a bad system — the error bar (±28.88) is too wide. AIVAT collapses it to ±11.56, so from just 34 trades the edge is decisively negative (~−$74.69/trade). Of the −$3,478.34 raw loss, −$938.85 was genuine bad luck; −$2,539.49 was systematic — and that systematic bleed ≈ fees (−$2,268.68).

Decision-leak verdict · net −$4,637.35 · fees −$2,268.68 (49% of the loss)

L3 no stop / ticket of record×38

L6O overtrading (>3 trades/day)×32

L6R revenge (entered into a 2-loss streak)×22

L1 oversize (notional beyond risk budget)×31

L5 churn (<30min hold or <0.5% move — fees eat it)×33

L8 averaging / re-entry after a losing close×12

L2 BTC/ETH correlated same-direction stacking×4

Single-cause attribution (each loss charged to its top-priority leak): L6R −$3,864.28 · L1 −$520.29 · L6O −$366.13 · L8 −$0.44

What discipline would have saved

All five rules enforced → −$472.73 instead of −$4,637.35 — saving $4,164.62. The disciplined month takes 3 trades, not 39.

Even then it's slightly negative — discipline caps the damage; positive edge needs a real system. First-order (scale/drop trades & re-sum): an estimate, not a guarantee.

R1 risk ≤ $150 (notional ≤ $30k)saves $2,104.30

R3 no churn (<30min / <0.5% move)saves $2,728.06

R2 no correlated BTC/ETH stackingsaves $901.11

R5 max 3/day · stop after 2 lossessaves $3,958.09

Worst decisions once luck is removed

Where raw P&L and AIVAT diverge, the outcome misled you. Some losses were mostly bad luck; some “small” losses were actually bad decisions that got a lucky path.

Trade	Raw P&L	AIVAT	luck
XTBRDOZ · BTC short	−$691.02	−$262.27	−$428.75
XTWDSSD · ETH long	−$393.57	−$242.54	−$151.03
XTNZPFO · ETH long	−$539.47	−$181.00	−$358.47
XTNPNDW · ETH short	−$204.41	−$155.70	−$48.71
XTYYUGV · ETH short	−$52.13	−$144.93	$92.80
XTJUCMI · ETH long	−$361.64	−$134.99	−$226.65
XTY3335 · BTC short	−$15.94	−$131.47	$115.53
XTKKOBE · BTC short	−$195.18	−$103.93	−$91.25

Replay a decision · candlestick hand review

The trade analogue of a poker hand review: step the slider to watch the price path unfold against your entry/exit, and compare the raw outcome to the luck-adjusted grade.

XTBRDOZ · BTC short · 32 min holdraw −$691.02 · AIVAT −$262.27

bar 33/33

at this bar · price 63027.7 · unrealised −158.0 px (short: up = loss)