Progress towards Model 5.6
Model 5.2, 5.3, 5.4, 5.5, 5.6 research and development progress update
2 years since last research update
It's been exactly two years since we wrote about Model 5. Since then, we have made great progress in improving the model's capability. Not just the model's performance, but also the production trading system, data pipeline, fallback handling, logging, fees, spreads, latency, and more. In particular, for our model, we have made rapid iterations with Model 5.2, 5.3, 5.4, 5.5, and 5.6. Each represents big changes while representing the original Model 5 architecture.
We won't be able to explain many details here, and some points will be ambiguous with not much explanation, but we'd like to still mention a few points on changes and improvements that have been made.
Improvements with Model 5 generation
From Model 5 to 5.2, we made improvements to reduce volatility and catch more opportunities. This change was made to allow us to scale further with more AUM.
Towards Model 5.3, we improved to avoid the downside of reduced profit in the previous change in Model 5.2 while capturing the same benefit of reduced volatility. Along the way, we made Model 5.3 freer to take risks for larger profits.
In Model 5.4, we introduced the capability to catch more market-neutral opportunities with equivalent risk level.
In Model 5.5, some of the downsides of previous iterations were changed and improved, and more focus was placed on calculated risk management, dynamic position handling, and the time range of trades. With this iteration, handling fees have improved.
Our latest model, Model 5.6, brings full multi-strategy capability. Model 5.6 brings and combines all the improvements that have been made upon till Model 5.5, as well as its original capability with improvements.
Model 5, in particular, the Model 5.6 version, which now powers Alpha Fund, deserves its own section to explain its vision and philosophy behind it. This will be brought up in the next post on competitive advantage.
Work left to do
Optimization
There is still so much work left to do. We will first continue operating with Model 5.6 and ensure everything goes right as expected. There are still optimizations we could do to further improve and ensure our live system follows exactly as we intended it to do.
Internal tools
All of our strategies are built in-house as proprietary strategies. Over time, as we have developed Model 5 generations towards Model 5.6, we have developed a lot of proprietary built-in tools. We will continue to improve in this aspect.
While significant progress has been made, we still have much to prove ourselves with Model 5.6. It's only been operated for a short period of time. We'll continue to remain focused on research, disciplined execution, and building systems that can perform consistently across changing market conditions.
Interested in our systematic approach?
View Model 5 Performance