Key takeaways
- Execution is judged by what repeats. One bad fill means nothing; the same fill every Tuesday does.
- Four signals matter: requote density, spread spikes, latency drift, and divergence from an outside feed.
- Median latency hides the damage. The slowest fill in twenty is the one that costs you.
- Spreads that widen at the same hour with no news are scheduled, not caused by the market.
- Log broker quotes beside a reference feed with the clocks aligned, or you cannot prove anything.
You judge a broker's execution by what repeats, not by what happens once. A single bad fill means nothing. The same bad fill every Tuesday at the same hour means a great deal.
So compare quotes, fill speed and spreads session by session, and split calm markets from wild ones.
First set a baseline, then decide how often is too often. That line is what separates a rough market from a broker whose execution is simply getting worse.
What Rigged Execution Looks Like
Requote Density Cluster
Frequent requotes in a calm market point two ways. Either the routing is shaky, or the broker is picking which orders to take.
It matters more when the same pairs, at the same hours, do it again and again.
Spread Spike Concentration
Spreads that widen at the same hour every day cost you twice. You pay more to trade, and your stops get hit more often.
Split the spikes that follow news from the ones that arrive on schedule with no news at all.
Latency Drift
In busy hours the slowest fills get slower. The price you get drifts further from the price you asked for.
Feed Divergence Persistence
If your broker's price keeps differing from an outside feed, its price source is either poor or inconsistent.
When the clocks line up and the gap is still there, it is worth digging much deeper.
Several signals going wrong together, again and again, tells you far more than one bad day.
Log your broker's prices next to an outside feed. Without that second source you cannot prove a thing.
Keep both the typical number and the worst one for each session. The first shows a normal day, the second shows a bad one.
Diagnostic Matrix
Measure every number over the same window, with the clocks lined up. Otherwise you are comparing two different things.
| Pattern | Observed Metric | Interpretation |
|---|---|---|
| Requotes | Requote rate by hour and instrument | Routing stability indicator |
| Spreads | Median and tail spread distribution | Cost-friction profile |
| Latency | Median vs p95 fill delay | Execution consistency marker |
| Feed Divergence | Quote delta vs reference feed | Price-source integrity marker |
| Stop Asymmetry | Stop-fill quality versus market-fill quality | Execution-control marker |
During the London overlap the spread holds steady and requotes are rare. The same pair at rollover widens again and again, and fills slow down with it.
Split the day into sessions and the cause shows itself.
Evidence Prioritization
- Trust what repeats. Ignore the one-off.
- For each number, keep the typical case and the worst case.
- Line up the clocks before you call a price gap real.
Analysis Discipline
Use the same window, the same clock and the same set of pairs every time. Change any of them and the runs stop being comparable.
Never edit the raw log. Build your summaries in separate files, so anyone can redo the work from the start.
Conclusion
To catch a platform bending prices you need clean data and a steady baseline. Then compare calm markets against wild ones.
Requotes, drifting spreads and a price feed that wanders are the three signals worth watching.
Any one of them can spike by chance. All three, again and again, is a case you can take to the broker.
This page explains how to read execution data. It gives no financial advice.
FAQ: Execution Patterns
What is a requote cluster?
A short stretch where the broker keeps replacing your order price far more often than normal.
Why compare p95 latency instead of only average latency?
The average hides the bad moments. The slowest fills are the ones that cost you.
How is feed divergence validated?
Line up the clocks. Put the broker price next to an outside feed and check whether the gap keeps coming back.
Why are recurrent session-specific spikes important?
A problem that returns at the same hour has a cause you can find. Random noise does not.
What improves confidence in execution-pattern conclusions?
All four signals moving the same way, over and over, in the same windows.
Does this article include financial advice?
No. It shows how to read execution data. It gives no financial advice.
Methodology Note
We log quotes and fills and split them by session. Then we line them up against an outside feed and count how often each pattern comes back.
It only works if the clocks match and the numbers mean the same thing every time. The same pairs have to be sampled in every market.
- A baseline per session for every number we track.
- The typical fill against the slowest one in twenty.
- Every price gap checked against an outside feed.
A live case of execution behavior under momentum conditions is documented in the Bullcharge live review.
Related Investigations
Continue with adjacent findings from the same research cluster.
Prop Firm Challenge Math
Rule-density and path-dependency analysis for challenge structures and payout feasibility.
Read Investigation→FCA's Two-Tier Justice
Retail leverage caps against a GBP 1m cartel settlement with no fine, no admission, and a firm in administration.
Read Investigation→Prediction Markets Casino
Event contracts at $50.6B a month, 85%+ sports volume, and two federal insider trading prosecutions.
Read Investigation→Offshore Brokers
Jurisdiction risk map for entity routing, enforcement distance, and payout protection gaps.
Read Investigation→