Paper trading only. No real money and no broker connection. Results depend on your cost settings.
Chart
Open model trades
| Opened | Market | Side | Model | Session | Entry | Stop | Target | Risk |
|---|
Decision log
| Bar | Market | Session | Model | Decision | Reason |
|---|
Profitable trades
Only closed paper trades that made money after costs.
| Closed | Source | Trade | Session | Size | Costs | Net | R |
|---|
All closed paper trades, winners and losers
| Closed | Source | Trade | Session | Exit | Net | R | Journal note |
|---|
Arbitrage scan
Loading.
Edge over 24 hours
Above zero means the route paid before fees. The dashed line is the fee you set, so the edge must rise above it to make money.
Why this is so hard
A three-leg loop pays the fee three times. At Kraken's published 0.20% per trade, the books must be out of line by about 0.60% before a loop breaks even.
Each scan reads the books one after another, so quotes can be seconds apart. A signal is only re-priced after the delay you set, and it is recorded whether it won or lost.
All routes, latest scan
Mid gap is the mispricing using mid prices. Edge uses the real bid and ask at that size. Thin markets show large negative numbers because their spreads are wide.
| Route | Type | Size | Mid gap | Edge before fees | Net | Best edge seen | Scans above zero | Quote gap |
|---|
Live quotes
| Pair | Venue | Bid | Ask | Spread | Bid size | Ask size | ECB reference | Updated |
|---|
Price-action models
The chart concepts, turned into fixed rules that can be tested. Every 5 minutes each closed 15-minute bar is checked against them.
Backtest
| Model | Market | Data | Trades | Won | Net R | Profit factor | First half | Second half | At double cost | t-stat | Tests |
|---|
A result passes with at least 30 trades, a profit factor of 1.3 or more, a profit in both halves, a profit at double cost and a t-statistic of at least 2. Backtests don't include the news window, because past calendar data isn't available.
Results by session
| Session | Backtest trades | Won | Net R | Avg R | Paper trades | Paper net |
|---|
Session clock
Hours from your session sheet, set in New York time and shown here in UK time, so they follow both clock changes.
This week's high and medium impact news
| When (UK) | Currency | Event | Impact | Forecast | Previous | Markets affected | Models |
|---|
Calendar from Forex Factory's weekly feed, refreshed hourly while trading is on.
Which news moves what
| News | Watch | Markets here |
|---|
The usual reaction to US data
Stronger jobs, inflation, sales or GDP, or a rate rise, tends to lift the dollar and weigh on gold. Weaker data tends to do the reverse.
That is a rule of thumb, not a signal. Prices react to the surprise against forecast, and to what was already priced in, so the models don't trade the direction of a release. They stay out around it.
Position size calculator
One standard lot is 100,000 units, or 100 ounces of gold. On your gold sheet, 1 pip is $0.01, so a 1.50 move is 150 pips.
Why the $20 to $50,000 ladder isn't a model
It risks about 23% of the account per level, to make 30%, and needs 30 wins in a row.
| Win rate per level | Chance of 30 straight wins | Chance of 3 straight losses |
|---|
Even at 70% wins, the chance of finishing is about 1 in 44,000. One loss also sends you back a level or more.
Gates before real money
The first four are measured on forward paper trades only. Backtests don't count.
Playbook
Your snapshots, what the portal does with each, and how far they can be relied on.
From snapshot to portal
| Snapshot | Used here as | Status |
|---|
Three strategies, one desk
- Triangular arbitrage. Three pairs on one venue, in a loop back to USD.
- Cross-venue arbitrage. Buy on one venue and sell on another. It needs money already held on both.
- Price action. Takes market risk, so it isn't arbitrage. It stands or falls on the tests.
What the evidence says
A study of second-by-second bank prices from 2005 found most triangular gaps lasted under 5 seconds, and about 94% were below 0.01%. It concluded they were very hard to capture after costs.
The Mirage of Triangular Arbitrage in the Spot Foreign Exchange Market. Markets have become faster since.
What isn't covered
- Real gold: XAUT, a gold-backed token on Kraken, stands in for XAU/USD and can drift from it.
- NZD and exotics: no usable free quotes, so they're left out.
- GBP/JPY: not listed on either venue, and USD/JPY spreads on Kraken are too wide to model.
- Forwards, futures, options and swaps: these are mainly hedging tools, with no data here.
Before any live step
- The 1 basis point model cost is an assumption until checked against a real broker's spreads.
- Prices come from Kraken and Bitstamp, not a bank or broker feed, so another venue's charts will differ.
- UK retail CFD leverage is capped, and trading from the UK needs proper legal advice.