How We Implemented the Arxum Trading Protocol: Backtesting to Live Results
Most trading systems fail in production. The backtest looks clean, you go live, and the first week hands you three consecutive losses you never saw in the data. We’ve been through that. This is the story of how we built our core trading protocol, what broke during testing, what we fixed, and what the live account numbers actually look like after running it on a $600 Exness account for 90 days.
For the full strategy reference, see the swing trading guide this protocol is built around.
What the Arxum protocol is
Not a magic indicator. Not a signal service. A structured decision framework that answers three questions before every trade:
- Is the market in a trending or ranging regime?
- Does the entry have confluence from at least two independent signals?
- Does the risk-to-reward justify the trade at current volatility?
If all three answers are yes, you enter. If any is no, you wait.
Simple to describe. Executing it consistently is harder, so we turned it into a repeatable process.
Phase 1: Building the rules
We started with a hypothesis: most retail losses come not from bad strategies but from ignoring context. Traders apply a trend-following approach in a ranging market, or a breakout system when price is compressing before a major news event.
The first version of the protocol used three regime filters:
- ADX(14) threshold: ADX below 20 = ranging, above 25 = trending. Between 20-25 = ambiguous, skip the trade.
- Higher timeframe structure: 4H must confirm the direction of any 1H entry.
- Session filter: no entries in the 30-minute window before major economic releases (EUR/USD, US NFP, Fed statements).
We coded this in Pine Script and ran it on TradingView across EUR/USD, BTC/USDT, and XAU/USD from January 2023 to December 2025.
Backtesting results (3-year data, 3 instruments)
| Instrument | Trades | Win Rate | Avg R:R | Net return | Max drawdown |
|---|---|---|---|---|---|
| EUR/USD | 312 | 54% | 1.8:1 | +34% | -8.2% |
| BTC/USDT | 187 | 49% | 2.4:1 | +41% | -14.1% |
| XAU/USD | 224 | 51% | 2.1:1 | +38% | -11.3% |
Win rate below 55% across all three. That’s intentional. The protocol is designed around R:R, not win rate. A 50% win rate with 2:1 average R:R produces a positive expected value of +0.5R per trade.
One pattern we noticed: the protocol underperformed badly in Q3 2024 across all instruments. That period coincided with very low volatility; ADX consistently below 20. The system was sitting out trades or entering barely-qualifying setups. This was a data point, not a flaw.
Phase 2: Forward testing on demo
We ran the protocol on a $10,000 demo account for 30 days before going live.
Results matched the backtest directionally but not numerically:
- Win rate: 47% (slightly below historical average)
- Average R:R achieved: 1.9:1 (slightly below 2.1 target)
- Biggest issue: slippage on fast-moving BTC entries; demo fill prices didn’t reflect real market conditions
The second issue was behavioral. Even on demo, we found ourselves second-guessing signals. Twice we skipped confirmed setups because the news calendar looked “risky.” Both turned into clean winners.
We added one rule from this phase: no override. If the three conditions are met, you enter. If you skip it, you log it as a missed trade in the journal. Accountability over discretion.
Phase 3: Going live on $600 Exness
We chose $600 because it hits Exness’s second CPA tier, where standard accounts get tighter spreads. FXStreet covers EUR/USD spread analysis if you want to compare broker conditions before picking your platform. At 1% risk per trade, each stop-loss represents $6, real enough to sting.
90-day live results (March 3 – May 2, 2026):
| Month | Trades | Winners | Losers | Net P&L | Drawdown |
|---|---|---|---|---|---|
| March | 22 | 11 | 11 | +$47 | -$42 |
| April | 19 | 10 | 9 | +$61 | -$38 |
| May (partial) | 7 | 4 | 3 | +$23 | -$19 |
| Total | 48 | 25 | 23 | +$131 | -$42 peak |
Win rate: 52%. Average R:R achieved: 1.85:1. Net return: +21.8% over 90 days.
The peak drawdown of $42 happened in week 2 of March: three consecutive losses during a low-ADX EUR/USD compression period. After tightening the ADX gate to 25 minimum for trend entries, the drawdown profile improved.
What we changed after going live
Three adjustments, all minor:
1. ADX gate raised from 20 to 25 for trend entries. The backtest used 20. Live trading showed too many borderline entries in the 20-25 range.
2. BTC position size capped at 0.8% risk (not 1%). Crypto volatility means the stop needs to be wider. 0.8% keeps the dollar risk comparable to Forex entries.
3. No trading XAU/USD during London–New York overlap (13:00–14:30 UTC). Gold spreads widen significantly during this window on Exness Standard. The extra slippage was eating into R:R on otherwise valid entries.
None of these changes invalidate the core logic. They’re execution refinements.
What the protocol doesn’t solve
Emotional management during losing streaks. Three losses in a row feels bad regardless of your system’s expectancy. The journal helps, but you can’t automate your way through a losing run.
It also does not adapt to black swan events. During those periods the protocol mostly sits in cash, low ADX, elevated volatility. That’s fine. It’s not built for crisis trading.
How to implement this yourself
The complete protocol: with Pine Script code, entry/exit rules, and the position sizing formula. Available to Telegram channel members. We update it when we make rule changes (always versioned, always documented).
Entry levels, stop losses, and lot sizes. Updated every trading day. Join free.
FAQ
Can beginners use this protocol?
What's the minimum account size?
What instruments does it work on?
What happens when the protocol produces conflicting signals?
🌍 Our recommended brokers
Some links on this page may earn us a commission — at no extra cost to you.
Reader Reviews
The three-question framework, trend or range, confluence from two signals, R:R justified, seems obvious when written out, but having it as a checklist rather than an internalized heuristic is completely different in practice. I was skipping the R:R check on 30-40% of my entries. Making it a mandatory gate on every trade cut my average losing trade size by about 25%.
The section on iterating the protocol after backtesting rather than abandoning it when the first results were below expectations is the most valuable part for me. I've thrown out three systems in the past year when they had short-term drawdowns. The ADX filter addition story, identifying a specific problem and fixing it without changing everything else, is a much better model for system development.
The behavioral issues section during live trading is something I've never seen discussed honestly in retail trading content. The observation that even with a profitable system people deviate from rules under pressure, and the specific solution of reviewing every deviation rather than just every losing trade, is genuinely useful and not something I've read elsewhere.
The +27.2% over 90 days result is credible because the article also shows the drawdowns and the trades that didn't work. A cherry-picked result would only show the good months. The 44-trade sample with 55% win rate and 2:1 average R:R matches the theoretical expectancy almost exactly, which tells me the reporting is accurate.
Valuable case study. The Pine Script code mention is interesting, it would be useful to have that published publicly so readers can run the same backtest on their own data and timeframes. The protocol itself is solid and well-documented. The one addition I'd suggest is a section on how to handle correlated positions when trading multiple instruments simultaneously.
The backtest methodology section is what made me trust the results. The explanation of why he backtested on out-of-sample data and avoided curve fitting, rather than optimizing on the same data he was testing, shows a level of rigor you don't see in retail trading content. Results presented this way are actually meaningful.
The section on position sizing evolution, starting with 0.5% risk per trade and only scaling to 1% after 50 verified trades with the system, is the most practical onboarding protocol for a new strategy that I've seen. Most traders jump to their target risk immediately. The patience required here is the actual edge.
The monthly P&L review process described at the end of the article is something I've now adapted for my own trading. The specific format, categorize each loss as system loss, execution error, or external event, then only address execution errors through rule changes, avoids the trap of over-optimizing after normal variance. Solid framework.
Leave a Review
Crypto Trader & Technical Analyst
Crypto trader since 2019. Specializes in momentum strategies using RSI, MACD, and volume analysis on Binance Futures. Has managed personal portfolios through multiple market cycles.