Skip to content
All strategies

Strategies

Guyamin FMZ BTC Spot Turtle Breakout

Outcome: Abandoned

GuyaminBtcSpotTurtleBreakout

Outcome Summary

This long-only BTC spot strategy used daily Donchian breakouts with ATR-based sizing and exits. Its backtest reported 694.50% return and a 1.11 Sharpe ratio across 37 long positions. After three iterations, it was abandoned at backtest review because returns appeared while flat and the performance history stopped before the reported end date; optimization was not reached.

Hypothesis

Faithful hypothesis based on Guyamin’s [海龟策略btc现货版](https://www.fmz.com/strategy/289308). Trade BTCUSDT.BINANCE_SPOT long-only on completed daily bars in a CASH account with no leverage. Published defaults: fresh_rete=24 hours, trade_percent=0.01, DC_range=30, atrlength=24. The author’s separate 2019-01-01 to 2020-03-02 backtest used daily bars and overrode DC_range=20 and atrlength=14. Its claimed 80% annualized return and 16% maximum drawdown are unverified.

The trading rules remain unchanged. Daily cash, marked BTC holdings, and fills are logged so a complete rerun can reconcile equity and pyramiding. The available report has 500 sampled equity points and eight closed positions but no fill ledger; the 36 trade events remain unconfirmed. The prior 30/24 run is the separate default-parameter baseline. The required outbox write failed because this session’s filesystem is read-only, so this JSON was not saved there.

suspected backtest artifact: recurring nonzero returns while flat and performance history ending seven months before the reported backtest end. Optimization would fit an unreliable equity curve.

Implementation

Guyamin BTC spot daily Turtle breakout using the published backtest’s 20-day channel and 14-day ATR settings.

Verification Results

The sandbox reports positive USD PnL and total return but negative average return on deployed notional. Reconcile the fill ledger and return calculation before using that field for fee analysis.

Backtest Review

Sharpe
1.11
Total return
694.50%
Max drawdown
39.99%
Trades
37
Win rate
37.8%
Profit factor
3.12

The 37 recorded positions are long-only, matching the spot breakout hypothesis.

Suspected backtest artifact: daily returns show −0.1969% and +0.1973% on 2018-09-27/28, when the trade list shows no open position.

The report says the backtest ended on 2024-06-19, but its daily returns, equity curve and recorded trades stop on 2023-11-10.

The prior iteration requested reconciliation of flat-day equity and the missing performance history; both discrepancies remain.

Analysis

Reconcile the missing performance history and verify flat-day equity against cash, positions, and fills. Confirm that the eight positions and 36 trade events reflect the intended pyramiding. Rerun the complete period before optimization. ## Library refinements (from the knowledge library; test them, do not assume them) The library supports reproducing Guyamin’s published 20-day channel and 14-day ATR settings before testing changes to pyramiding, entry filtering, or exits. The required outbox write was rejected……Show moreShow less

Reconcile the missing performance history and verify flat-day equity against cash, positions, and fills. Confirm that the eight positions and 36 trade events reflect the intended pyramiding. Rerun the complete period before optimization. ## Library refinements (from the knowledge library; test them, do not assume them) The library supports reproducing Guyamin’s published 20-day channel and 14-day ATR settings before testing changes to pyramiding, entry filtering, or exits. The required outbox write was rejected because this session’s filesystem is read-only, so this JSON was not saved to the requested file. 1. [parameter] Reproduce the published settings: Run a separate BTCUSDT.BINANCE_SPOT daily-bar baseline with DC_range=20, atrlength=14, fresh_rete=24 hours, and trade_percent=0.01. Keep 30/24 as a separately labeled baseline. Reconcile daily equity as cash plus marked BTC holdings, match all eight positions to their entry, add, and exit fills, then rerun the complete period before comparing results. — Guyamin’s published backtest used 20/14, while this run uses 30/24. The missing equity history and unresolved relationship between eight positions and 36 trade events prevent a reliable interpretation of its reported return and drawdown. (source: 海龟策略btc现货版 p.1) 2. [sizing] Cap pyramid units: As a separate variant, allow one initial buy and at most three adds. Add only when a completed daily close exceeds the previous filled buy price by 0.5 × ATR(24). Size each unit as the smaller of 1% of account equity divided by ATR(24) and available spot cash divided by the execution price with a 0.2% cash buffer. Count cash and marked BTC holdings exactly once in account equity. — Guyamin specifies 0.5 ATR adds and equity-based volatility sizing without a unit cap; the library’s TurtleRules implementation caps pyramiding at four units. This variant tests whether bounded exposure reduces the observed 49.8% drawdown. (source: 海龟策略btc现货版 p.1; TurtleRules p.1) 3. [filter] Filter buys by the long trend: As a separate variant, permit a 30-day Donchian breakout buy or pyramid add only when the completed daily close is above the SMA of the prior 200 completed daily closes. Leave exits unconditional and require 200 prior closes before trading. — A library Donchian strategy uses SMA200 to confirm long breakouts and identifies sideways-market false signals as a risk. This may improve entry quality, though the current eight-position sample is too small to predict its effect. (source: 基于唐奇安通道和移动平均线的趋势突破交易策略 | Trend Breakout Trading Strategy Based on Donchian Channel and Moving Average p.1) 4. [exit] Test a ratcheting ATR exit: As a separate exit variant, retain the 15-day lower-channel close exit. Initialize a long-only trail at the entry fill minus 3 × ATR(10); after each later completed daily bar, raise it to the greater of its previous value and that close minus 3 × ATR(10). Exit the full position when a completed close falls below the previous bar’s trail. Keep Guyamin’s last-buy-price minus 2 × ATR(24) exit in the faithful baseline. — The library’s Donchian strategy uses a 10-period ATR with a 3 ATR trailing exit. A trail that cannot fall may limit drawdown, but it could also cut the large winning trades behind this run’s high profit factor. (source: 唐奇安通道突破策略|Donchian Channel Breakout Strategy with ATRSL Trailing Stop p.1; 海龟策略btc现货版 p.1)

Iteration History

Verification failed (Layer 4 — QA review) [class=code_defect]: - [edge_concern] The sandbox result covers only 11 closed trades; it does not establish the full-history edge. - [critical] _last_buy_price resets to zero after a restart with a carried BTC position. Warmup does not restore it, so half-ATR adds and the required two-ATR exit remain disabled. (line 22) — fix: Restore the last filled buy price before processing live bars, using persisted fill state or fill history. - [warning] _last_buy_price changes on order submission, even if the buy is rejected or never fills. (line 94) — fix: Update it only after a confirmed fill.

Outcome Summary

Reconcile flat-day returns and the full performance history before optimizing a backtest.

After three iterations, the backtest-review verdict was abandon because returns appeared while flat and recorded performance ended months before the reported backtest end. Optimization and later reviews were not reached.

It traded long-only BTC spot breakouts on daily bars using Donchian levels and ATR-based sizing and exits.

The backtest reported 694.50% total return, a 1.11 Sharpe ratio, 37 long positions, and 39.99% maximum drawdown. The performance record was found unreliable.

Analysis

海龟策略btc现货版

Backtest and paper results are hypothetical. Trading involves risk of loss.