Strategy Audit: an ETH Averaging Scalp That Wins 96% of Its Trades — and Still Loses Money
A real audit of a real client strategy, published verbatim (with permission) as the sample of our paid strategy audit. Every paid audit follows this exact structure: the claim, the replication, the numbers, the failure mechanism, and the limits of what we measured.
The strategy's 94–98% win rate is real. Its profits are not: after exchange fees and one asymmetric stop, every meaningful test window ends net-negative. The advertised +467%…+1,538% per year could not be reproduced under honest conditions.
1 · What the client brought us
A live ETH grid scalper: averaging entries, +0.5% take-profit per cycle, 5% stop, direction picked by a hand-built WaveTrend reversal indicator, running both directions at 4× leverage. A third-party results table claimed +467% to +1,538% per year with "no losing year". We treated that table as an unverified claim — that is the default in every audit.
2 · How we tested it
- Real ETH candles, multiple timeframes and windows — never a single lucky month.
- Strict anti-look-ahead: signals use only bars that were closed at decision time.
- Pessimistic intrabar rule: when a candle touches both stop and take, the stop fills first.
- Real costs: 5 bps per side, no fee rebate, no free fills.
- A causal reimplementation of the WaveTrend direction filter, compared against a plain EMA baseline.
3 · What we measured
| Window | Direction filter | 胜率 | Net return |
|---|---|---|---|
| 30m candles · 7 months | EMA baseline | ~96% | −10.7% |
| 30m candles · 7 months | WaveTrend | ~96% | −6.2% |
| 15m candles · 1 month | WaveTrend | ~95% | −5.9% |
| 5m candles · 2 weeks | WaveTrend | ~98% | slightly + |
The WaveTrend filter genuinely helps — it roughly halves the losses versus the EMA baseline. It just never turns the system positive. The only positive window is the shortest one: classic regime luck, not edge.
4 · Why a 96% win rate loses money
Each take-profit earns +0.5%, but the round-trip fee is ~0.1% — the exchange takes 20% of every winner off the top. Meanwhile one −5% stop erases roughly ten winners at once. High win rate + tiny takes + rare large stop is the classic martingale-scalp illusion: the equity curve climbs in small steps and falls in an elevator.
The client's own live bot confirmed it in one day: take-profits +$1.52, stops −$10.11, net −$8.59 — balance below its starting deposit after peaking earlier.
5 · Limits of this audit — stated, not hidden
6 · What NOT checking costs
From the same lab, same methodology, before anyone risked live money:
7 · Get this report for your strategy
The price is fixed before you send anything — no meter, no surprise invoice. You get the same structure as above: claim, replication, numbers, failure mechanism (or a genuine pass), limits.
One strategy, one market, one honest backtest against fees and a matched baseline. Written verdict like the sample below, within days.
Multi-window walk-forward, cost model, null baselines, overfit checks, and a fix-list. Fixed quote agreed before we start.
Deep-dive on your rules or portfolio: parameter search on our compute, forward paper book, ongoing verdict updates. Maximum quote up front.
Check my strategy
Describe your rules in the request form — we quote a fixed price, then run the audit. A negative verdict is a valid result: it is the cheapest loss you will ever take.
Check my strategy →Submitting requires a free account. This sample stays public — no signup to read it.
Also live today: bot unlock $20 one-time · Terminal Pro $25 / 30 days · free honest scoreboard on what we sell.