Live AI Trading Bot Testing Results: Session Archive

Account Overview (May 3, 2026)

OKX sub-account dashboard. Starting deposit on March 10, 2026 visible as the vertical step in the equity curve. Estimated total value as of May 3, 2026: $5,497.35 USDT. Sub-account UID and name shown for verification.

Performance Snapshot

How to read these numbers: The 100.00% win rate counts only closed trades. 22 positions remain open with −$111.46 USDT unrealized loss — those positions are not yet in the win rate denominator. The bot's strategy holds losing positions open while taking profit on winning ones, which is why floating losses can persist for weeks while closed-trade win rate stays high. Both numbers must be read together. Max drawdown of −$83.32 USDT (~1.67% of starting capital) is the worst equity dip experienced during the session — useful as a stress measure independent of current state.

OKX Performance Stats (Futures, all-time). 926 closed positions, +$609.20 total profit, 100% win rate, average hold time 1 day 5 hours. The +24.73% average PnL% is calculated per-trade against position margin — it is not the account return. Average PnL per trade is +$0.66 USDT.

Floating Positions Detail

22 open positions across multiple symbols (ZRO, CHZ, SAND, FIL and others). Each position has a floating loss visible in the Floating PnL column. Total floating PnL: −$111.46 USDT. This is the side of the bot's hedge currently against the market — these positions will close in profit when the market reverses, which is the design assumption of the dual-direction approach. Until they do, the loss sits on the books.

Maximum Drawdown — March 29, 2026

Lowest equity point during the session: $4,916.68 USDT on March 29, 2026 — about 22 days into the test. This represents a maximum drawdown of −$83.32 USDT, or 1.67% of starting capital. The dip occurred during the bearish March phase. From this low, equity has recovered through April's sideways-to-bullish conditions and May's continued strength.

Profit Guard — Realized Profit Transfer

On April 29, 2026, this account's Profit Guard threshold triggered. Profit Guard is a CoinTech2u feature that, when configured, automatically transfers a fixed amount from the trading account to the funding wallet each time realized profits exceed a set equity-growth threshold. This account is configured to move $250 USDT to funding on every 5% equity growth. The transfer isolates locked profits from active trading risk — once $250 sits in funding, it cannot be drawn down by future market moves on the trading side. The $250 is included in our session net return calculation but no longer participates in trading.

Monthly Trading Calendars

OKX provides a per-month trading calendar showing daily closed P&L and closed position count. These give a direct visual of how performance varied across the three market regimes during this session.

March 2026 — Bearish phase

March (24 days bot ran): +$280.16 USDT total closed P&L. The session started March 7. BTC was in a bearish phase during this window — short positions closed in profit while long positions accumulated as floating exposure (driving the March 29 drawdown).

April 2026 — Sideways to bullish

April: +$317.41 USDT total closed P&L — the strongest month of the session so far. Sideways price action allowed both sides of the hedge to take profits on pullbacks, and the late-April uptrend helped the long side dominate. This is the regime the bot's strategy is best matched to.

May 2026 — Continued bullish

May 1–3: +$11.62 USDT closed. Session continues running through the bullish phase.

What 57 Days Taught Us

The first three weeks of this session ran through bearish conditions. As BTC declined through March, short positions closed in profit while long positions accumulated as open exposure — driving floating P&L into negative territory and producing our worst equity dip at $4,916.68 on March 29, a $83.32 drawdown from the $5,000 starting capital. April flipped the tape: sideways price action with frequent pullbacks let the long positions take profit on upward swings, and both sides of the hedge contributed to closed gains. By late April, BTC began trending up, the long side dominated profits, and Profit Guard triggered its first $250 transfer on April 29.

The pattern across 57 days is that markets did not stay in one direction for long. Bearish weeks gave way to sideways weeks gave way to bullish weeks. A discretionary trader who positioned for any one of these regimes would have been wrong about the next one — and would have spent the entire window watching the screen to find out which way it had shifted.

The bot is not predicting which regime is next. It runs both sides of the hedge continuously, takes small profits as they appear, and absorbs floating losses on the side currently against the market. That is a mechanical edge, not a forecasting edge — and the result over 57 days is that the account moved from $5,000 to $5,497.35 (plus $250 isolated in funding) without any human discretion required. Whether that mechanical edge holds during a more violent regime — a multi-week one-directional crash, or a flash-crash event — is the question the next phase of testing has to answer.

“

The bot is not predicting which regime is next. It runs both sides of the hedge continuously, takes small profits as they appear, and absorbs floating losses on the side currently against the market.