Something I noticed in my log last month that I hadn't fully clocked before: when I trade US30 in the afternoon, I'm often holding something correlated with it in the same hour without thinking of it as related.
The obvious one is NAS100. I rarely trade both at once, but it happens, and on a risk-off move they fall together.
Less obvious is USDJPY. On many afternoons, especially when US yields are moving, USDJPY and US indices move in the same direction. If I'm long US30 and long USDJPY, I'm doubling up on a "risk on, dollar up" view whether I meant to or not.
I went through September. On 11 afternoons I had a US30 position open at the same time as at least one other position. On seven of those, the other position was in a correlated instrument moving the same way. My worst afternoon of the month was one of those seven, when a sudden drop at 15:10 London hit both a US30 long and a USDJPY long within the same minute.
So now there's a rule: if I hold US30, any other position in that hour counts as part of the same risk budget. Combined risk across them can't exceed what I'd risk on a single US30 trade.
It means fewer trades. I can live with that.
Correlation caution: US30 plus related risk in same hour
-
LondonScalper
- Posts: 965
- Joined: Sat Sep 05, 2026 7:54 am
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Correlation caution: US30 plus related risk in same hour
Counting the hour as one risk budget is the right fix. The attached chart is an illustration of the kind of afternoon you describe: US30 and USDJPY both drifting up from 14:30, then both dropping inside the same minute at 15:10.
Where I'd push it further is the correlation itself. It isn't fixed. On days when the move is driven by yields, USDJPY and US30 can go opposite ways, so a blanket rule might be stricter than needed on some afternoons and still too loose on others.
What I do is check the last 60 M1 bars before adding the second position. If both instruments moved the same way on most of those bars, it's one trade for budget purposes. If not, I size them separately.
Your 7 out of 11 is a useful number though. It says that most of the time the second position wasn't diversification, just more of the same view.
Did the combined rule cut your number of afternoon trades much, or mostly just the size of the second one?
Where I'd push it further is the correlation itself. It isn't fixed. On days when the move is driven by yields, USDJPY and US30 can go opposite ways, so a blanket rule might be stricter than needed on some afternoons and still too loose on others.
What I do is check the last 60 M1 bars before adding the second position. If both instruments moved the same way on most of those bars, it's one trade for budget purposes. If not, I size them separately.
Your 7 out of 11 is a useful number though. It says that most of the time the second position wasn't diversification, just more of the same view.
Did the combined rule cut your number of afternoon trades much, or mostly just the size of the second one?
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Correlation caution: US30 plus related risk in same hour
Thinking more about your 15:10 drop, I'd check what was on the calendar that day. Plenty of afternoons have a 15:00 London release such as ISM, consumer confidence or JOLTS, and a move ten minutes later is often the second leg of it.
If it was, then the shared risk was the event rather than the correlation. US30 and USDJPY both react to the same US number, whatever their usual relationship happens to be that week.
So I've added a simpler rule on top of the 60 bar check. If there's a US release between 15:00 and 15:30, every index and USD position open in that window counts as one trade.
The other thing I learned the hard way is that two stops hit in the same minute both slip. On one fast drop my combined loss was about 1.4 times the planned combined risk, because each fill was worse than its own stop.
That's why I'd set the combined budget a bit under a single US30 trade rather than equal to it. Did that afternoon in September line up with a scheduled number, or was it a headline?
If it was, then the shared risk was the event rather than the correlation. US30 and USDJPY both react to the same US number, whatever their usual relationship happens to be that week.
So I've added a simpler rule on top of the 60 bar check. If there's a US release between 15:00 and 15:30, every index and USD position open in that window counts as one trade.
The other thing I learned the hard way is that two stops hit in the same minute both slip. On one fast drop my combined loss was about 1.4 times the planned combined risk, because each fill was worse than its own stop.
That's why I'd set the combined budget a bit under a single US30 trade rather than equal to it. Did that afternoon in September line up with a scheduled number, or was it a headline?
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Correlation caution: US30 plus related risk in same hour
Tick volume counts how many times the price changed in a bar. It does not count how many lots traded. In a decentralised market like spot FX, that is all most retail platforms can offer.
The surprising part is that tick volume and real volume correlate fairly well on liquid pairs. Studies comparing the two on major currencies have found a decent relationship, which is why many traders use tick volume as a rough activity gauge.
The weak spot is the exceptions. A burst of quote updates from one liquidity provider can inflate tick counts without any real trading. Different brokers produce different tick counts for the same minute. And during quiet periods, tick volume can look flat while a single large order moves the market.
The sensible use is relative, not absolute. Compare the current bar to the typical bar at the same time of day on the same platform. A bar with twice the usual tick count is worth noticing, whatever the number means in lots.
One way to calibrate is to compare tick volume on the platform with CME futures volume for the same hour over a few weeks. On liquid pairs the two usually move together, and the correlation figure gives a sense of how much trust the tick count deserves on that particular broker.
Anomalies are informative too. A day when tick volume spikes but futures volume stays flat suggests quote noise from the feed, not genuine participation. A day when futures volume jumps but tick volume barely moves suggests the broker's feed is filtering heavily. Either way, the comparison turns a vague indicator into something with known limits, and known limits are what make an indicator usable.
Has tick volume ever given you a useful warning, or does it just add noise to your charts?
The surprising part is that tick volume and real volume correlate fairly well on liquid pairs. Studies comparing the two on major currencies have found a decent relationship, which is why many traders use tick volume as a rough activity gauge.
The weak spot is the exceptions. A burst of quote updates from one liquidity provider can inflate tick counts without any real trading. Different brokers produce different tick counts for the same minute. And during quiet periods, tick volume can look flat while a single large order moves the market.
The sensible use is relative, not absolute. Compare the current bar to the typical bar at the same time of day on the same platform. A bar with twice the usual tick count is worth noticing, whatever the number means in lots.
One way to calibrate is to compare tick volume on the platform with CME futures volume for the same hour over a few weeks. On liquid pairs the two usually move together, and the correlation figure gives a sense of how much trust the tick count deserves on that particular broker.
Anomalies are informative too. A day when tick volume spikes but futures volume stays flat suggests quote noise from the feed, not genuine participation. A day when futures volume jumps but tick volume barely moves suggests the broker's feed is filtering heavily. Either way, the comparison turns a vague indicator into something with known limits, and known limits are what make an indicator usable.
Has tick volume ever given you a useful warning, or does it just add noise to your charts?