Our strategy SETH CRUDE was released on June ’22 and since then are not having profits.
Since launched, and for the previous reference sizing of 1.2 contracts, there is an accrued total of -130€ approx. Not a successful story but without a doubt not bad (non-solid algos just start to have heavy losses after releasing it)
The last December 2022 and January 2023, for example, were not good, and the point here is that having a plain curve of profits is not our goal (a plain curve of profits is not the same that a stagnation period, but in this case, the last months are not suggesting a promising future)
Coincidentally, we found that SETH CRUDE is working fine with TF 10 min, and our checks seem that it could continue going well. Current subscribers can just change the TF launch backtest to see it.
And even better: while we were adapting SETH CRUDE to IBKR (that means when adapting the strategy from using the CFD of IG Markets, to using the standard futures of Interactive Brokers), we found enhancements for the strategy that applies in the same way when we use TF 10 min.
We were waiting for an Out Of Sample period, but in fact, were a few of options and variations, and we didn’t launch a specific version in demo or real to show you.
We mean that given the version adapted to IBKR (and not released) and using TF 10 min, the results are so good, but we didn’t notice it until now, while we were revisiting the development versions.
We feel that we have to release this new version, in TF10 min, just because the current SETH CRUDE is going nowhere and our analysis makes us think that the new version V1.6 using TF 10 min will be significantly improved (you just have to see the next comparisons between the previous and the new version)
Then, what are finally the new V1.6 release changes compared to the previous V1.4.2?
- Change to TF 10min and a simple variation and optimization of some parameters
- Include launch long positions only with half sizing (a big part of profits come from short positions, even in uptrends)
- Adds new controls to avoid the system stopping on market data flow empty values (that raise a division by zero error)
This is the final result (1.75 contracts, similar height to 1 NASDAQ contract, with a spread 3. BT for 200k candles)
Here is the same configuration, but launching the max history available (1M candles) — Note that adding more than two years of time and around 50% more operations (from +500 to +750), the ratios and averages are similar. That’s a good signal too.
The good point and key in our decision to publish this new version now, is that the adaptations of this V1.6 are in fact to low. If we get the previous V1.4.2 version in the last 200k in 1min (near the OutOfSample period of the V1.4.2), and we launch the same version in TF 10min….the enhancement is considerable and the period of sample is near 7 months.
On next you can see the OutOfSample period of the unreleased version V1.5, which we used now to create the V1.6. Using TF10, the results are awesome. And to adapt V1.5 to V1.6 we just verified that the long positions work better with the first entry only
Then, what changes have this new V1.6 vs the non-released V1.5?
The V1.5 version vs V1.4.2 have a simple variation and optimization of some parameters, but that can affect significatively to the performance and the ‘version consistency’. But V1.5 was developed on 7 Nov and is when looking at results with TF 10min, when we see great behaviour in TF 10min in the OOS period.
Now, to release V1.6, we just changed the behaviour of longs (launching only half position) because in the long term have better results and considering that does not affect the strategy (is in fact a filter of already validated OOS results) and applying a new fix to avoid possible market data problems
Do you wanna see yourself this strategy? Download our demo for backtest with this and all our systems
Do you like this new version? Subscribe to this strategy (only at 89€/3 months or 269€/year, with 25% discount)