Puedes leer este artículo en español, haciendo click aquí
If your DARWIN is growing in assets under management, congratulations: you're doing things right. But that growth brings with it a challenge that many traders overlook until it starts affecting their investors' returns: the divergence and its impact on the ability.
This guide contains the most practical advice so you can diagnose your current situation and take concrete steps before the problem escalates.
A DARWIN's capacity is the maximum investment volume it can absorb before order execution begins to deteriorate noticeably. As you grow, the volume of orders that Darwinex replicates for your investors increases, and this higher volume can generate two symptoms:
The divergence is calculated as:
Divergence % = (investor price − strategy price) / strategy price × 1000 (multiplied by −1 if it is a purchase order)
For example: if your buy order is executed at 1.15384 and the investor replica is executed at 1.15389, the divergence is 0.00433%.
Problems typically arise when order sizes exceed a nominal value of 1 million. Above this threshold, negative divergence can begin to systematically erode returns for your investors.
Before taking action, you need to understand where the problem lies. The section: Investor Divergence from your main account.
For a complete description of all sections and metrics, please refer to the Official Darwinex documentation on investor divergence Here's a summary of the essentials.
1. Info DARWIN It displays the overall parameters: target VaR, VaR ratio, invested capital, number of investors, cumulative monthly divergence, and average latency. It's a dashboard at a glance. If the monthly divergence is negative and becomes significant (above 0.2% per month), it indicates that there's work to be done.
2. Strategy and investors' orders This is the most comprehensive section for conducting an analysis. For each order you execute, an aggregate order is generated for all your investors. Here you can see them side-by-side:
The four graphics They are especially useful:
Practical tip: Download the complete CSV data (select "By date range", up to 12 months of historical data) and cross-reference it with your trading times. If you see that orders executed at 9:00:00, 10:00:00, or 14:30:00 have consistently worse divergences, you have a clear pattern to correct.
3. Net strategy and investors' positions It displays real-time net open positions: yours, DARWIN's, and investors', along with their respective VWAP (volume-weighted average price). If any symbol appears in red, there is an inconsistency between your position and investors' positions that requires immediate attention.
Each investor order can be divided into deals (fragments that Darwinex sends to the liquidity provider) and in turn in fills (the partial execution of each deal). The field Div. VWA pip It tells you how much each fill contributes to the total divergence of the order.
Example An order of 58.33 lots with a total divergence of 8.43 pips can be broken down into:
If the first deal accounts for almost all of the divergence, the problem lies in the size of the first partial execution, not the entire order. This gives you very specific information about what to adjust.
This is the most common mistake in algorithmic strategies and one of the most impactful. If your algorithm is set to enter at the exact beginning of a candlestick, you're competing with thousands of orders from other traders doing the exact same thing at that very moment.
Why does it get worse as the timeframe increases?
Because more traders use those candlesticks as a reference. A 1-minute candlestick opening only attracts those who use that timeframe. A 1-hour candlestick opening attracts everyone: traders using 1-hour, 30-minute, 15-minute, 5-minute charts...
The solution: Introduce a delay into your algorithm's logic. The recommended starting point is wait 5 seconds after the candle starts before placing the order.
Concrete example If your system operates on 1-hour candles and the buy signal is confirmed at the close of the 2:00 PM candle, schedule the execution for 2:00:05 PM instead of 2:00:00 PM. A small change that can significantly reduce divergence in those orders.
Note: For assets linked to futures (indices, commodities), sensitivity is lower and 1-2 seconds may be sufficient. For FX, 5 seconds is more relevant.
Darwinex has a tool for toxicity which answers the question: what performance would my DARWIN have obtained if it had entered X seconds later in each operation?
The tool generates a curve that shows the optimal delay for your specific strategy, for both opening and closing orders.
Example of interpretation If the toxicity curve shows that entering 5 seconds later would have improved performance by 0.3% monthly, you could implement that change, improving capacity and profitability. However, if the curve drops rapidly (the delay would cause you to lose many pips of the signal), fractional trading might be a better strategy than delaying.
These charts are available in the Capacity attribute of your DARWIN.
Order splitting involves dividing a large order into several smaller ones, sent with a slight time lag between them. By reducing the impact on the order book, both execution efficiency and divergence for investors are improved.
How to do it without increasing leverage?
This is the critical point. Fractioning only makes sense if you can divide the lot without changing the overall risk of the trade.
If you're in the unfavorable situation, the solution involves Increase your account equity until you have room to split your trades. For example, if you increase your account balance from €10,000 to €20,000, your reference lot size increases proportionally by two times, and you can then divide it.
Where to begin: opening or closing?
In most cases, divergence problems are concentrated in the openings because that's where the trader (especially the algorithmic one) competes most aggressively for price. However, from a practical point of view, it's usually easier to implement fractional trading in the close first, because it does not involve changing the signal input logic.
Even if you only split the closing price, the divergence may improve, but you will have to analyze your specific case.
The trade-off you need to know
More fractional trading generally means better divergence for your investors, but it can also mean slight losses in signal efficiency (you enter or exit at prices marginally different from your backtest). There's an optimal balance that every trader must find for their strategy. The goal isn't to maximize capacity at any cost: if fractional trading reduces your annual return from 10% to 8%, but triples your capacity, it might be worthwhile. If it drops you to 6%, it probably isn't.
Latency isn't just a technical metric: it's a barometer of your order flow. Low, stable latency indicates that your trades are reaching the market during periods of normal liquidity. High latency or latency with recurring spikes indicates the opposite.
Manual traders typically have very low latency because they tend to trade when the price is stable, during periods of lower activity. Algorithmic traders, who chase price movements, usually have higher latency because they enter the market just as it is most active and the trade opening signal is generated.
Warning signs in latency:
If you see high latency but no significant divergence yet, don't get complacent: the divergence could be circumstantial (perhaps the last 100 orders were favorable by chance). Latency is the leading indicator.
Not all markets react the same way to volume. There is a relevant difference between:
Futures-linked assets: indices, commodities, stocks and ETFs.The reference price is the futures contract or the quoted security, a centralized and highly liquid market. A 1-2 second delay is usually sufficient to avoid congestion at the beginning of each candlestick.
FX: When trading against prime brokers that offer slightly different feeds, execution is more sensitive to the exact timing. Here, the 5-second delay is more relevant, and the toxicity tool tends to show more pronounced differences between entering "on point" versus with a delay.
All the above theory is only valid if you compare it with your actual data. The downloadable CSV file from the Investor Divergence section (available by date range, up to 12 months) contains, for each order:
With this data, you can perform a simple analysis in Excel or with the help of AI, seeking answers to questions such as:
Each of these questions has a specific action associated with it: delay, splitting, changing operating hours, or reviewing the problematic symbol.
If you've arrived here and don't know where to start, this is the recommended order:
For a complete guide to all metrics available on the platform, see the Darwinex's official article on the Investor Divergence section.
Thanks for reading,
The Darwinex Team
*Trading involves risk. The content of this article is for informative purposes only and is not to be construed as financial and/or investment advice.