When to Trade EURUSD Iâve found an interesting time related effect in the EURUSD and Iâm interested if anyone else has noticed. The issue is time and specifically the start and end point for creating bars. This particular system trades 30 min bars with the maximum trade length of 2 bars or 60 minutes. Entry is at the market and exit at either a stop loss, profit target or in 60 minutes. My starting point is 1 min bars data which I use to create 30 minute bars. The big difference comes in the point I use as the 0 minute for the 30 inute bar. The standard practice would be to use the hours 0 and 30 minute marker as the starting point, so the bar at 0:30 minutes begins with prices after 59 minutes and before 30 minutes of the hour. When I ran a test using 15 and 45 minutes as the bar start minute, instead of standard 0 and 30 I noticed a radical change in performance. 30 min bar start at 0 and 30 minutes: 5.5 PIPS 30 min bar start at 15 and 45 minutes : 10.7 PIPS The same system with the same logic showed an almost 100% increase in the average trade just by changing when to start creating 30 minute bars. Has anyone noticed this kind of change or done any research in this area? Jerry
Now that's something I've not heard of before! How far back have you gone to test this theory to make sure it's not just coincidence, and what price are you taking, the close of the previous bar or the open of the 15 and 45? Why not just use a 15m bar chart?
It's not even a theory at this point...just an observation. I've only looked at a few hundred trades, so as you note it requires further measurement over more history. I'm taking the profit of the same logical system: applied to 00 start minute bars and 15 start min bars. The profit is measured from entry at bar open to exit at close of the bar 60 minutes after entry...market entry and exit. A possible theory at this point is that since most trading applications default to bars starting a minute 00, be they 15 min, 30, 60 or whatever, people make trades and decisions just after a new bar is available. So there is a "rush" of trading at 1 or 2 minutes after each new bar. This front loads standard bars to some degree with information on price changes. Bars created at a 15 minute offset form time 00 shows this differently in terms of bar open to bar high, etc. Since my system looks at bar patterns it picks up signals in the 15 min start bars that arenât shown in the 00 start bars. Actually the best approach may be to trade both sets of bars as the results of both are profitable and only about 65% of the signals overlap.
I'd often wondered myself if there wasn't an artificially created bias due to this anchoring in clock time. Just like there are "magic" numbers in price (e.g. decade numbers) maybe many systems are triggering nearly simultaneously based on bar time.
It makes sense, I imagine a lot of traders/systems wait for a close to confirm a trade (in order to avoid spikes maybe?). I trade off a 15m bar chart and often notice a 'lull' just before the close of a bar, if that's of any significance.
The attachment in the prior post shows the difference in range based on the 30 minute starting point: minute 00 vs. min 15. The 00 min bars tend to have higher ranges, which could alter the bar patterns and explain my difference in results.