GhostRider

Overview

Multi-timeframe DCA futures strategy. It finds support (Launchpad) and resistance (Firewall) zones on 15m candles, checks the trend on 4h and 1d, then trades long or short with a capped number of DCA adds. Version 4.05 adds a stop-loss set as a share of margin (the Balanced preset closes a position after losing 50% of its margin), entries when the trend turns, and a hard cap on DCA adds. Partial sells, wick entries and trailing stops are optional.

• Trades long or short with the trend
• Stop-loss on by default: closes at 50% of margin lost
• DCA with a hard cap on adds per position
• Enters when the trend turns
• 15m entries with 4h and 1d trend checks

What's new in 4.05

Released 2026-10-12. This is the first release since 4.01. It includes the changes from internal builds 4.02 to 4.04.

This page covers GhostRider for TradeStaq bots. For the GhostRider strategy for Gunbot, see GhostRider (Gunbot).

Fixes

• More reliable trend reading. Direction and entry decisions now follow the trend as intended. In 4.01 the trend could be misread.
• Targets and stops no longer change with leverage. Profit targets, stop-losses and margin checks are now measured on the price move. A 2% profit target means a 2% price move at any leverage. Before, the leveraged return was used, so raising leverage quietly tightened every target.
• Trend-confirmed stop. With SL_TREND_CONFIRM on, the stop now also fires when the trend is strongly against the position. Before, a strong counter-trend blocked the stop, which is exactly when it was needed.
• Strict confirmation works both ways. The stoch_strict confirmation option now confirms short setups as well as long ones.
• Entries in strong trends. Opportunistic first entries now also open when the trend is strong, as intended.

New options

• Margin-based stop-loss. With SL_MODE set to MARGIN, SL_VAL is the most you are willing to lose as a percentage of the position's margin. The price stop is SL_VAL divided by leverage, so a 50% stop at 10x closes the position after about a 5% price move against it. Balanced uses 50% and Conservative uses 18%.
• Trend-flip entries. With ENTRY_ON_TREND_FLIP on, a flat bot opens one trend-following position when the trend turns bullish (long) or bearish (short), without waiting for a support or resistance zone. Balanced turns this on.
• DCA cap. MAX_DCA_LAYERS is a hard limit on how many times a position can be added to after the first entry, whatever the zone logic says. The default is 5 and Balanced uses 7.
• Stop timeframe. SL_ATR_TF chooses which timeframe's volatility (ATR) sizes an ATR or volatility stop.
• Short fallback in auto mode. With AUTO_FALLBACK_FOLLOW_TREND on, an auto-mode bot that is flat, with no zone choosing a side, falls back to a short in a bearish trend instead of always going long.

Recommended setup

Run GhostRider on 15m candles with 4h and 1d as its two trend timeframes, using the Balanced preset. The backtests below used this setup, and so do the live GhostRider bots.

The stop-loss is off in the strategy's own defaults. If you build your own settings, turn it on. Without a stop, a DCA position that keeps going against you can lose a large part of the account. In backtests (Bybit perpetuals, 15m with 4h and 1d, 10x leverage, 10% size, in-sample, not live results), 4.01 without a stop lost an entire ZEC test account between 2025-06-15 and 2025-12-15, and 4.05 without a stop still lost 47% on HYPE between 2026-04-08 and 2026-10-08. Methodology & evidence has the full setup. Backtested results do not indicate future results and are not financial advice.

New and changed settings

Presets

Balanced and Conservative both have trailing stops off and use 15m candles. Conservative also waits for a stronger trend before entering and uses the strict confirmation option. Aggressive has no stop-loss and is not offered. Only Balanced has been backtested in the tests below. Position size is the share of your balance used for each entry's value. Margin is that value divided by leverage.

Requirements

Required Indicators

Configuration Parameters

Multi-Timeframe Configuration

Recommended Position Sizing

Matches the Balanced preset: 10% position size at 10x, with the stop-loss at 50% of margin.

Version Info

Current Version: v4.0.6

Changelog

Methodology & evidence

Backtested, hypothetical results, not live trading. Every figure below comes from simulations on historical data. They are in-sample: the 4.05 Balanced settings were chosen while looking at this same data. Past or simulated performance does not indicate future results. Leveraged futures can lose more than the margin on a position, and a DCA strategy can lose a whole account. Nothing on this page is financial advice.

How the tests were run

• Exchange and market: Bybit USDT perpetual futures, using 15m candles with 4h and 1d as the two trend timeframes.
• Pairs: BTC, ETH, HYPE, ZEC, ADA and SUI. Each pair ran separately on its own $10,000 account in cross margin. "Avg net" is the average return of those six accounts.
• Size: 10x leverage. Each entry and each DCA add was worth 10% of the account balance, so it used about 1% of the balance as margin.
• Costs: 0.05% commission plus 0.05% slippage on every fill. Funding payments were not modelled.
• Settings compared: 4.01 Balanced, the preset as it was before this release, with no stop-loss, against 4.05 Balanced, which adds the 50%-of-margin stop, trend-flip entries and a cap of 7 DCA adds. All other Balanced settings were the same in both.
• Code: 4.01 and 4.05 rows used the exact code of each release. Positions still open at the end of a window were closed at the last price, and that loss counts in the result.

"Worst max DD" is the deepest peak-to-trough fall of any single pair's account.

Four 6-month windows: 4.01 against 4.05 with the stop

Experiment GR-2026-10-12-v405-windows. Windows C and D overlap from 2026-04-08 to 2026-06-15.

This is a trade-off, not a free improvement. Across the four windows, 4.05 with the stop earned less on average than 4.01 without one: −0.2% against +7.8% per pair per window. Most of that gap comes from window A, where 4.01 did well and the stop closed 248 positions for a combined loss of $18,166 across the six accounts. In exchange, the worst result for any pair improved from −100% (a wiped-out account) to −10.5%, and the worst drawdown from 100% to 13.4%. 4.05 finished positive in 10 of the 24 pair-windows, against 18 of 24 for 4.01.

The stop limits losses but does not cap them exactly. In these tests the worst single 4.05 position lost 1.28 times its margin.

In windows C and D, much of 4.01's loss came from DCA positions still underwater when the window ended: $3,706 and $7,348 across the six accounts. 4.05 had $82 and $18.

The ZEC wipeout (4.01, window B)

On 2025-09-04, 4.01 opened a ZEC short near $40.54. As ZEC kept rising, it added to the short six times, the last near $75.16. The position was liquidated on 2025-10-01 and the $10,000 ZEC test account went to zero. ZEC rose 816.5% over that window. Over the same window, 4.05 with the stop finished ZEC at −2.1% with a 7.8% maximum drawdown.

Results by pair

Net return per window, 4.05 Balanced first and 4.01 Balanced second.

Buying and holding these coins over the same windows ranged from −53.8% (ADA, window C) to +816.5% (ZEC, window B). GhostRider trades both directions with DCA and is not designed to track buy and hold.

Recent window: what the fixes do without the stop

Experiment GR-2026-10-12-defaults-recent. Window D only (2026-04-08 to 2026-10-08), same setup, using the strategy's built-in defaults rather than a preset.

The 4.02–4.05 fixes lifted the average in this window, but on their own they did not stop HYPE from building a large losing DCA position. The stop did. This is why Balanced turns it on, and why we recommend keeping it on.

Results last recomputed on 2026-10-12 from the raw backtest output. Backtested and hypothetical: past or simulated performance is not indicative of future results. Not financial advice.