ForumSections

How do you measure the real impact of economic news on an EA? - page 3

New comment
Den Martin #:

One detail on the slippage-setting point: if an EA passes that input to MqlTradeRequest.deviation, its effect depends on the execution mode. That field is used with Instant/Request execution and ignored with Market/Exchange execution. Lowering it therefore doesn't cap slippage on a Market Execution symbol. I'd record the execution mode alongside requested and filled prices when comparing news trades.

MetaQuotes documents the distinction here: https://www.mql5.com/en/book/automation/experts/experts_market_buy_sell


Hi Den,

thanks, I didnt know that. So a "max slippage" input can look like protection in the settings and do nothing at all on a market execution symbol.

If I buy an EA with such an input, I'd now ask the seller which execution mode it was tested on. And the requested vs filled price on a few live releases shows quickly if the setting does anything or not.

Best regards,
Chris
Chris K. #:
Hello Youness,

yes, the calendar in the tester was the reason i started asking sellers about it.

Your last point is the one I'd check first. A filter that only blocks new entries still leaves every open trade sitting through the release, and thats exactly where the stop gets filled far away from its price. So "has a news filter" tells me very little until I know which kind it is.

Best regards,
Chris
Hello Chris,

Agreed. When I ask a seller about it, I ask three things: does it only block new entries or also close or protect open trades, how many minutes before and after the release, and where the news times come from in the tester. If they can't answer the third one, the backtest is the "without" version.

For open trades there is also a middle option between "do nothing" and "close everything": move the stop to break-even or take partial profit a few minutes before high-impact releases. It keeps the trade but limits what a bad fill can cost.

Best regards
Youness
Youness Kanbaou #:
Hello Chris,

Agreed. When I ask a seller about it, I ask three things: does it only block new entries or also close or protect open trades, how many minutes before and after the release, and where the news times come from in the tester. If they can't answer the third one, the backtest is the "without" version.

For open trades there is also a middle option between "do nothing" and "close everything": move the stop to break-even or take partial profit a few minutes before high-impact releases. It keeps the trade but limits what a bad fill can cost.

Best regards
Youness
Thanks Youness,

I'll add the third question to my list. If the seller cant say where the news times come from, that already answers it.

On the middle option I'd split it in two. Moving the stop to break-even doesnt change much on a real release, because a stop at entry gets skipped the same way as any other stop. Taking partial profit is different. A smaller position is what actually limits the bad fill.

Best regards
Chris
Christian Kotschote #:
Thanks Youness,

I'll add the third question to my list. If the seller cant say where the news times come from, that already answers it.

On the middle option I'd split it in two. Moving the stop to break-even doesnt change much on a real release, because a stop at entry gets skipped the same way as any other stop. Taking partial profit is different. A smaller position is what actually limits the bad fill.

Best regards
Chris
Hello Chris,

Glad the third question is useful. And fair split: break-even is still just a stop, so a real release can skip it like any other. The smaller position is what actually limits the fill.

There is also a second problem with break-even right before news: on many brokers the spread widens before the release. A stop sitting at entry can then be hit by the spread alone, before price has really moved. So you pay the wide spread, lose the trade, and sometimes watch it go your way afterwards.

So if break-even is used at all, it needs room for the widened spread, or it's better done well before the event, not at the last minute.

Best regards
Youness

The first few minutes of news broadcasts can cause significant slippage, which is extremely unfriendly to most types of EAs.

You might be opening a position and stopping out at a very bad price, far worse than you set your entry point.

i tested this on a gold breakout EA, only for NFP. ran 7 versions of the block window, from 5 min before / 2 after up to 120 before /
  60 after, about 2.5 years of tick data. pretty much no difference, PF 1.49 vs 1.48, and the wider windows actually missed one of the
  big trend days. so for that system the filter did nothing useful and i left it off by default

  so what i'd measure is not the event itself but what your EA does inside the window. take only the trades the filter would block and
  look at their pnl on their own. if they are not worse than normal trades, the block is just costing you trades

  2 things that bit me. dont detect releases from price spikes, it kept mixing up CPI and NFP, use the real calendar. and dates move
  around, NFP wasnt always on a friday last year and one month didnt come out at all, so "first friday" logic breaks
@Yang Yu Qun that's a very useful test, and measuring only the trades the filter would block is the right way to look at it.

One thing I'd keep in mind when reading those numbers: with the tester's default "zero latency" setting, a stop is filled at the price in the tick history, with no execution delay. In the first seconds of NFP a live account can get filled later and worse than that. So the blocked trades probably look a bit better in the backtest than they would live. "No difference" is a good sign, but it's closer to a best case.

Two cheap ways to check it: rerun the same test with one of the tester's delay settings instead of zero latency, and on a demo or small live account log the requested vs actual fill price for the trades inside the window. If the gap stays small on your broker, leaving the filter off makes sense.

@Rui Nan Yang agreed, and it's worst for pending stop entries, since the breakout itself triggers them at the bad price.

Best regards
Youness
123
New comment