Who got through the last volatile weeks without strange fills?

[FXS London Scalper]

The last few weeks had a couple of violent sessions, gold especially, but also yen and the dollar around US data. I'd like to compare how different setups held up.

Not looking for broker bashing, just a factual picture. If you can, mention:

• Broker and entity (or at least type: ECN-style, market maker, hybrid).
• Platform.
• Instrument and time of the worst fill.
• How far your fill was from the stop or requested price.
• Whether the platform stayed responsive.

My own experience: on EURUSD nothing unusual, maximum stop slippage of 0.6 pips during US data. On gold it was worse: one stop filled about 40 cents beyond the level, and the platform was sluggish for maybe 20 seconds. No requotes, no rejected orders.

What I'm trying to figure out is whether the bad fills I've seen reported here are broker-specific or just what the market looked like at that moment. If several brokers show the same slippage at the same timestamp, it's probably the market. If one stands out, that's useful information.

Post your numbers, even if they're boring. Boring results are data too.

[Shadow Trader]

Last look is the part of liquidity provision nobody advertises. A provider streams a quote, a trade request arrives, and the provider gets a short window to accept or reject it. During that window, if price moved against the provider, the trade can be bounced.

For a scalper, this shows up as rejections or requotes mainly when price is moving in the direction of the order. Take the offer on a rising market and the trade is likelier to be turned down, because the quote is already stale from the provider's point of view.

Some venues publish a last-look policy, and some brokers let it be disclosed on request. Others route through multiple providers with different policies, so the behavior changes depending on who is on the other end that second.

Not every rejection is malicious. A provider protecting itself from latency arbitrage is acting rationally. But the practical effect is the same, and it skews results in a way that backtests never capture.

Could a broker ever explain its last-look setup in plain terms, and was the answer worth anything?