Volta V1.3 Execution Log

Runs

A run is a period during which the settings and the timeframe did not change. The moment either moves, the run ends and a new one begins — so a 1-minute run and a 4-hour run are never averaged into one figure. P&L is computed from the broker's own deals inside each window, so a run cannot drift from the account.

Every round trip, newest first

Reconstructed from the fills, not from the broker's statement. The strategy reverses rather than flattening, so one message usually closes a trade and opens the next; points are index points, and P&L is MetaTrader's own figure for the trade including commission and swap. Running is the cumulative total up to and including that trade, so with the newest at the top it counts down toward the start of the test.

Best +386.0 points, worst -84.6. Far too few trades to mean anything about the strategy — they are here because execution accuracy is only worth reading next to what it cost.

Every message, newest first

Times are the bridge's. “Sent” is what the script put on the wire; “placed” is what reached the broker. A reversal is one message that closes and re-opens, so it shows two fills and two outcomes. Stops are re-anchored to the broker's price at the moment of the order, which is why one lands within about a point of the distance asked for rather than exactly on it. The small grey number under each outcome is MetaTrader's own code, kept so a failure can be looked up.

The settings behind every number here

Taken from the script's source with changed values marked, because no entry has yet reported its settings. This can drift out of date; a reported set cannot.

Broker and instrument

Trailing Step is in chart ticks, not points: 1000 ticks on this 0.1-tick feed is 100 index points. The same input on a 0.01-tick feed would be ten points, which is the kind of thing that looks fine and trails ten times too tightly.

What the trades offered, and what was kept

For each round trip: the best it was ever worth (MFE), the worst it was ever worth (MAE), and the difference between the best and what was actually banked. P&L says whether a trade worked; these say why. Measured from one-minute bars between entry and exit.

Could a faster ratchet recover any of it?

Give-back being large does not mean it is recoverable. A tighter trailing stop keeps more of the trades that ran and cuts short the trades that would have run further, and which effect wins depends on the route each trade took — something MFE and MAE cannot show, because they are only its endpoints. So the real one-minute path of every trade was replayed against a stop sitting a fixed distance behind the best price seen.

The same figures, for anything that reads rather than looks: metrics.json — published with every rebuild, and carrying its own sample-size and accuracy caveats so they cannot be separated from the numbers.

Which conditions the strategy has traded in

Every session is classified from the 20 daily bars that closed before it opened — never from its own bar. That constraint is what makes the label usable: a condition you can only identify with hindsight is not something a strategy could ever have adapted to. Each round trip is then attributed to the condition that was already knowable when its day began.

Efficiency is how much of the market’s movement went somewhere: the net move over 20 sessions divided by the distance actually travelled. Near 1.0 every session pushed the same way; near 0 the market covered a great deal of ground and arrived back where it started.

What this data does and does not support

Written from the logs by the assistant running the test, and regenerated with them. The distinction it draws is deliberate: execution quality is answerable after a handful of orders, because it asks whether each message survived the trip. Profitability is not, and pretending otherwise would be the most useful-looking and least useful thing this page could do.

How a signal becomes an order

Five hops. The script owns the first one and nothing else; everything after it is replaceable, which is the whole point of the Middleman setting.

Three ways to run it yourself

You do not need the bridge. It is a test rig built to prove the script, not a product, and the script was written so it could be swapped out.

What the bridge does that a hosted service does not

What this run has and hasn't shown