How to read the hedge router dashboard

This page explains the dashboard in plain language. It covers what each panel shows, how much money you need to run the system, and how many normal clients it can serve. All numbers come from a computer simulation of one year, run 300 times. They are not real trading results.

The story in one paragraph

Your clients want to trade gold, silver, copper, Brent oil and crypto every day, including weekends. Your liquidity provider (LP) closes commodities from Friday 22:00 to Sunday 22:00 UTC. When clients trade with you, you take the other side, so if the price moves against you, you lose money. The simulation compares three ways to handle this:

A. Naive B-book: You keep all client trades and never hedge. On weekends you show the old Friday price for commodities.
B. Hedge when open: Like A, but you pass your risk to the LP whenever the LP is open. On weekends commodities stay unhedged.
C. Unified router: The new design. It sets limits per symbol, hedges on CME futures that trade on weekends, closes risk on Friday, buys option insurance, and charges a wider spread when the hedge is worse.

Result: A earns the most on average ($6.69M a year), but in a bad year it earns only $2.07M, and one bad weekend can cost $668k. C earns a little less ($5.51M), but even in a bad year it earns $5.08M, and its worst weekend costs only about $9k. C gives up about $1.2M of average profit to remove almost all of the big-loss risk.

The dashboard also shows D. Zero weekend risk: every trade is hedged at once, and symbols with no open hedge market are closed on weekends. It earns $3.38M a year and never has a losing weekend. Read the step-by-step zero-risk plan →

The panels, one by one

Panels are listed in the same order as they appear on the dashboard, from top to bottom.

Top tiles (six boxes)

The key numbers. Mean annual P&L is the average yearly profit over the 300 simulated years. 5th–95th pct is the range for 90% of years: only 1 year in 20 is worse than the first number. Weekend P&L per year is the profit from weekends only. Worst weekend is the biggest weekend loss in each simulated year, averaged over the 300 years. Hedging cost is what you pay the LP and CME to pass risk away. Sharpe is profit divided by how much profit jumps around, so a higher number means a smoother year.

Strategy comparison (table)

The three strategies side by side. Read across a row for one strategy. Important columns: Max drawdown is the biggest drop from a high point. Stale-quote loss is money lost when you show an old price on the weekend and informed clients trade against you (A loses $3.12M a year this way, C loses almost nothing). Rejected volume is client orders refused because they would push your risk above the limit. C rejects $1.63B a year. That is the price of safety.

Showcase year: cumulative P&L

One example year. The line shows your total profit growing day by day. Grey bands are weekends. Look at week 32, a forced "bad news weekend" where crypto falls 15–25% and oil jumps 12% while the LP is closed. A (purple) and B (orange) drop sharply. C (blue) barely moves.

Where the money comes from and goes

Bars above zero are income and bars below zero are costs, averaged per year. Spread revenue is your main income, the markup on every client trade. Market P&L is gains or losses from holding client risk. Hedge P&L and hedge cost are what hedging earns and costs. Option premium / payoff is the insurance you pay for and what it pays back. The last bar is the net result.

Monte Carlo band

All 300 simulated years together. The line is the middle (median) year. The shaded area covers 90% of years. A wide band means the result is uncertain (A). A narrow band means the result is predictable (C).

Weekend P&L distribution

A histogram of every weekend in every simulated year. Each bar counts how many weekends ended with that profit or loss (in $50k steps). C's weekends sit in a tall spike just above zero. A's and B's spread far to the left, into losses of $1M or more.

Risk desk — Friday close

What the dealer sees at 21:00 UTC on Friday of the stress week, one row per symbol. Client pos is how much clients hold against you. Hedge is how much you have passed to the LP or CME. Unhedged is the risk you still carry over the weekend. Weekend cap is the most risk allowed on that symbol during the weekend. Option cover is the insurance bought. VaR (value at risk) is the likely worst loss if the price gaps by the "gap assumed" amount. Use the buttons to switch strategy. For C, the unhedged column is zero after the Friday sweep.

Stress weekend, hour by hour

A zoom into the stress weekend for one symbol (use the buttons). Top chart: the solid line is the price you show clients and the dashed line is the true market price. In A, the commodity price freezes on Friday (flat line) while the true price moves away. Clients who know the true price trade against you. In C, the shown price follows the true price using CME futures or WTI oil. Bottom chart: your unhedged risk. In A it grows big over the weekend. In C it stays small.

By symbol (two bar charts)

Profit per symbol per year, then weekend profit per symbol. Crypto earns most because it has the most client flow. For gold (XAU), silver (XAG) and Brent, A and B lose money on weekends because of frozen prices, while C earns money there. Copper (HG) is "close only" in C on weekends, so it earns little on weekends but also loses little.

Router configuration (table)

The settings used for strategy C, one row per symbol: volatility, markup, which hedge is used on weekends (for example CME 1-oz gold), how much that hedge can absorb (capacity), how the symbol is quoted on weekends, how much wider the weekend spread is (for example 2.2×), and the weekday and weekend risk limits (caps). Changing this table changes the whole system. Nothing else in the code is symbol-specific.

How the simulation works

A short list of the assumptions: how prices and client orders were generated and how the hedging rules work. The most important point is that the data is simulated. Run the system on your own MT5 deal history before you trust the exact numbers.

How much money you need (strategy C)

The simulation assumes $5M of broker risk capital. Below, that number is broken into its parts, so you can see where the money is used and change the assumptions. This is your own money. Client deposits are separate and are not counted here.

Margin on crypto hedges (% of position) Margin on gold / silver / copper / oil hedges (%) Months of hedge & option costs to keep in cash Loss reserve for a failed weekend ($)

What the money is for	How it is calculated	Amount
Margin for crypto hedges	Crypto weekday caps total $5.2M (BTC 2.0, ETH 1.5, SOL 0.8, XRP 0.6, TON 0.3) × margin %
Margin for commodity hedges	Commodity caps total $4.3M (gold 2.0, silver 0.8, copper 0.5, Brent 1.0) × margin %
Running costs in cash	Hedge cost $2.82M/yr + options $0.15M/yr, for the chosen number of months. Spread income pays this back over time.
Loss reserve	Cash for a weekend where hedging fails. $1M ≈ the 1-in-20 worst week of strategy B (−$990k).
Total own capital

Why 50% for crypto: CME crypto futures and crypto exchanges ask for high margin, and if you hedge with spot crypto you pay 100%. Why 10% for commodities: typical LP and CME margin on metals and oil is about 5–10%. Ask your LP and FCM for the exact rates and enter them above.

Short answer: with these assumptions you need about of your own capital, which matches the $5M the simulation uses. In the simulation, strategy C returns about $5.5M a year on that capital. Even the 1-in-20 bad year returns about $5.1M.

Not included above (get quotes for these)

Regulatory minimum capital for your broker licence (depends on the country).
MetaTrader 5 server licence, Gateway and Manager API access, and hosting.
Minimum deposits at your LP and at the FCM (the futures broker for CME/ICE).
Staff (a dealer or risk desk on weekends), market data fees, legal and audit.
Marketing to win clients.

How many normal clients the system can handle

The simulation does not count people. It models the total net order flow from all clients together. Over a year, strategy C accepted $8.27B of net client flow (crypto $4.79B, gold $1.64B, Brent $0.90B, silver $0.58B, copper $0.35B) and refused $1.63B because of the risk limits.

Net vs gross: if 100 clients each buy $10k of BTC and 90 clients each sell $10k, gross volume is $1.9M, but your net risk is only $100k. Real gross volume is several times the net flow. The "gross-to-net ratio" below sets how many times.

Average trade size ($) Trades per active client per month Gross-to-net ratio

Active clients the current limits can serve

The table shows active clients for other trade sizes and ratios, using your "trades per month" value. "Active" means a client who trades every month. Registered accounts are usually several times more.

What limits the number of clients

The risk limits (caps), not the software. MT5 is not the bottleneck at this size. When more clients trade in the same direction, more orders hit the caps and get refused. Today about 16% of flow is refused ($1.63B of $9.9B).
Doubling clients raises net flow by 1.4× (if clients trade independently) to 2× (if they all follow the same news). To keep refusals at today's level, raise the caps by the same factor. Hedge margin grows by that factor too.
Weekend hedge capacity is fixed. CME 100-oz silver absorbs only about $200k and 10-bbl WTI about $300k. Weekend silver and oil cannot grow past these limits, whatever capital you add.

Best next step: export one year of MT5 deal history (hourly net client flow per symbol). Re-running the backtest on it gives exact numbers for your real clients instead of the estimates above.

Word list

P&L: Profit and loss.
B-book: You take the other side of client trades yourself instead of passing them to the market.
Hedge: A trade you make at the LP or CME that cancels your risk from client trades.
LP: Liquidity provider: the bank or firm that gives you prices and takes your hedges.
FCM: Futures commission merchant: your broker for CME and ICE futures and options.
Cap: The most net risk allowed on one symbol. Orders above it are refused.
Stale / frozen quote: An old price shown when the real market is closed. Informed clients use it against you.
Proxy: A similar product used when the real one is closed, for example WTI oil for Brent.
Strangle: An option pair (one call and one put) that pays if the price moves a lot in either direction. Used here as weekend insurance.
Friday sweep: At 21:00 UTC on Friday, the router hedges all open risk before the LP closes.
VaR: Value at risk: an estimate of a bad-case loss.
Percentile (5th, 95th): The 5th percentile is the result that only 5% of simulated years were worse than.
Monte Carlo: Running the same simulation many times with random prices to see the range of outcomes.

All figures come from a simulated backtest (300 paths of 52 weeks). They are not live trading results or investment advice.