Status

Checking... Looking for the userscript in this browser.

1. Install Tampermonkey

Tampermonkey is the free browser extension that runs the MetaDataOptimizer userscript.

2. Let it work on these three sites

Open the details of the Tampermonkey extension: paste the first line below in the address bar. Set Site access to On specific sites, add the three sites, and turn on Allow user scripts.

3. Install the userscript

Open the userscript: Tampermonkey shows an install page, press Install.

If that page does not offer Install: press Copy script code, open Tampermonkey, choose Create a new script, select everything, paste and save with Ctrl+S.

4. Prepare TradingView

1. Sign in to TradingView and save one layout per strategy (its symbol, timeframe and tools).
2. With the layout open, copy the code from the address: tradingview.com/chart/XXXXXXXX/. The code is the XXXXXXXX part. The whole address also works.
3. Create the alert of each family in TradingView; its name is what you write in the strategy.

5. Tell each strategy where it lives

Strategies → edit a strategy → Connections tab: paste the layout (link or code) and the name of its TradingView alert. Alert and Test strategies of a family can share the same alert.

6. Use it

1. Chart button (Strategies → Scripts): confirm the symbol, timeframe and inputs. TradingView opens that strategy's layout, puts the script on the chart, shows it in the Pine editor and saves the layout.
2. Bell button (only on the last two versions): updates the TradingView alert of the strategy to that version, with the same values. It restarts the alert.
3. While it works the TradingView screen is locked and says so; it unlocks by itself and tells you when it is done.

Good to know

The userscript updates through Tampermonkey (Utilities → check for userscript updates). If TradingView changes something inside, a step can stop working: the message on the panel in the corner of TradingView says which one.

What is 1R

1R is what you risk on a single trade: the distance between the entry and the stop loss. With a capital of 1000 and a risk of 2% per trade, 1R = 20. A stop loss costs 1R; a take profit of 5x earns 5R.

How often you must win to break even

Win one trade of TP×R and lose the rest at −1R: you break even when the share of winners is 1 ÷ (TP + 1). Commissions raise the bar. The bars below use 0.3R of commission per trade.

A higher target needs fewer winners, but winners also get rarer. The Reports tell you which side your data is on.

Commissions, also in R

Each trade pays its fees on the whole position, not on the risk. In R that is fee % × leverage ÷ risk %. Example: 0.09% × 6 ÷ 2% = 0.27R per trade. A tight stop means more leverage for the same risk, so the same fee weighs more in R.

Max R

Max R is the furthest a trade went in its favour, measured in R. It is how the app knows which take profit each trade could have reached. A trade only counts for a take profit of 8x if its Max R is 8 or more.

Where the numbers come from

The Alert records every operation, winners and losers. Your rules pick some of them. The Test records only the operations the rules accepted. The reports read both.

Dependence on big winners

What it answers. Does the result rest on a handful of big operations?

What it shows. The Net with and without the operations whose target is the multiple you set (10x by default), how much of the Net the best 5 and the best 10 winners make, and the list of the 10 best operations.

How to read it. If the best 10 winners make less than 30% of the Net the concentration is low (green); between 30% and 50% it is moderate (blue); above 50% it is high (red). The Net without the big ones should still be clearly positive.

Watch out. Big winners are the rarest. A result that only survives with them is a bet on seeing them again.

Thin conditions

What it answers. Which conditions took only one or two operations?

What it shows. How many conditions took a single operation and how many took the number you set or fewer, the Net resting on them, and the 20 thinnest conditions with their group, target and Net.

How to read it. A condition is identified by its group and its target. The fewer operations behind it, the likelier it is a coincidence of this data. The share of the Net backed by thin conditions tells you how fragile the whole set is.

Watch out. Conditions built one by one to catch single winners are thin by construction. That is not a mistake, but expect fewer operations in the future than in the history.

Group solidity

What it answers. For each group, are there enough winners at my target, and does taking the group pay?

What it shows. Reads the Alert of the same line (every operation, losers included). Per group: operations, winners that reached the target, losers per winner, whether that is below the break-even after commissions, the level of evidence (anecdote, weak, acceptable, solid) and the months left to reach enough evidence.

How to read it. Pays means fewer losers per winner than the break-even for your target and commission. Solid needs 8 or more winners and an accuracy clearly above the break-even rate. Months to solid use the pace the group has had so far.

Watch out. A winner is an operation that reached the target; everything else counts as lost. Groups with one or two winners tell you almost nothing.

Month by month

What it answers. Does the result repeat from month to month, or does one month carry it?

What it shows. Operations, TP, SL, Net and share of the Net of each month, the number of months with a negative Net, the best and worst month, and the Net without the best month.

How to read it. A steady result shows similar months. If the best month makes more than half of the Net (red) the history leans on a single month.

Watch out. The latest month is usually incomplete; do not read it as a bad or a good month.

After commissions

What it answers. What is left of the Net once every operation pays its fees?

What it shows. The Net before and after commissions, the commissions as a share of the Net, the same figure without the big winners, and the average Net per operation.

How to read it. Commissions are charged on every operation, winners and losers. If the Net after commissions without the big winners is still positive, the base of the result is healthy.

Watch out. 1R is read from the winners themselves, so it follows the strategy's risk and capital. The commission per operation is a flat figure in R that you can change.

How much can be loosened

What it answers. How many losers can a group let in before the profit starts to fall?

What it shows. Per group: the operations the rules left out, admitted one by one from the most similar to the accepted ones to the least similar. It shows how many to admit, how many losers that lets in, the gain at the peak and a reading.

How to read it. Loosen only when the peak has at least 2 extra winners and is more than one step wide. Otherwise it is a blip, most likely luck. With rules built on the same data there is usually nothing left to loosen: it becomes useful as new months arrive.

Watch out. The order is by similarity of the indicators, not by the exact limits of your conditions. It guides; it does not replace checking the condition itself.

Overlap between strategies

What it answers. When were two strategies open at the same time, and how far apart were their prices?

What it shows. Pick the Side (Buy or Sell) and two Test strategies of that side, then a month. The calendar paints each day; clicking a day opens its timeline with one bar per operation of each strategy, the stretches where both were open highlighted, and a table with the price at the start and at the end of each shared stretch.

How to read the calendar. Green both strategies were open at the same time · Blue only one of them had operations · Amber both had operations but never together · Empty no operations.

How to read the day. The day you clicked is in the middle; the day before and the day after are dimmed and only accompany it, so an operation that crosses midnight is not cut. Strategy A has a blue edge and B a purple one; the colour of the bar tells TP or SL. The amber band marks where both were open. In the table, Price at the start is the entry of the operation that opened later, Price at the end is the exit of the one that closed first, and Between entries is the distance between both entries in points and in %.

Watch out. Times are UTC. The SL and TP prices are not sent by the Pine: they are worked out from the entry, the leverage and the target. Only operations that carry the close date, close time, entry price and max exit price appear, so older months sent before those fields existed show as empty. A strategy can not be picked on both sides: choosing it in one selector removes it from the other.

Summary

What it answers. How is everything doing, in one screen?

What it shows. The key table of each of the six reports, with a link to open the full one.

How to read it. Use it once a month, after the new data arrives. Anything that turned red or blue deserves a look at its full report.

Watch out. It uses the filters of every report at once; set them first.

Colours

Green good or within the limit · Blue within the tolerance you set · Red beyond it.

Read it with this in mind

The rules were built looking at the same data the reports use, so every figure leans optimistic: a Test with no losing trade is the symptom. The numbers become more trustworthy with each new month the rules have not seen. Run the reports once a month and compare.

The idea: find the peak, not the maximum

Loosening a condition lets in more winners, but also more losers. The profit rises while each step brings fewer losers per winner than the break-even, peaks, then falls. Stop at the peak. Stopping short keeps a margin.

Steps

1. Start where there is volume. Work the groups with the most winners first; leave the thin ones alone until they have data.
2. Build the condition from one or two indicators. More filters fit the past better and the future worse.
3. Check the control columns. Net Fail, TP Lost and Expensive should be green, or within the tolerance you decided on purpose.
4. Count the evidence. A condition needs about 8 winners at the target before you trust it (Reports → block 3).
5. Loosen only with a clear peak. A peak that is one step wide, or has fewer than 2 extra winners, is luck (Reports → block 6).
6. Leave a margin. Take a little less than the peak.
7. Validate on a month the rule has not seen, and rerun the reports monthly.

How much evidence is enough

What the tolerance is for

Tolerance Net is how many losing operations a condition may let through. It starts at 0. Raise it when the curve shows that admitting those losers captures enough winners to pay: fewer than about 6 losers per winner at a target of 8x with typical commissions, and fewer than 3.7 at 5x.