ROA/M backtest results using the 1 week update interval. This backtest starts in early 2021 and ends just this week, Notice how the default strat (the top one) performs very well over all market conditions. However in the past year as the market has been more stable the -1 strat has really shined more than doubling the USD value of the principle position, but while the market was super volatile peak bull and bear run -1 was relatively stable not growing but maintain value. Starting at default and then switching to -1 when the market stabilized would have resulted in around 400% APY in 2.5 years, not including CAKE or CUBE rewards. We are going to use this data and make some adjustments to the range setting options. The first thing that comes to mind is making the adjustment between the settings a little more subtle. We will have the update interval adjustment results soon, as well as more insight on the machine learning algo we are using to create these results. Check back testing results on twitter ⬛️🚀 I've been holding off on public statements until I have major updates. As some of you saw from the tweet we made 2 days ago and as I've been explaining in the DAO meetings, we had been seeing deviations in share allocations to liquidity providers who are adding the same value to our shared position. We knew this is not right and had been continuing testing every day for many hours a day for weeks trying to find the source of this deviation. As people who have been paying attention to the space know, this deviation is exactly what caused Kyber to be hacked a couple weeks ago. This continued testing with out a firm solution is why we have just been keeping our heads down and continuing to work until we find a solution. Over the weekend we decided to stop testing our share allocation model and create a new one that we know will be accurate since we built it from scratch rather than going with what some other projects have done. This commitment to testing and security is why we have never suffered a major exploit or hack. It would have been very easy for us to say, 'well this is what all the other projects are doing so it must be right', or 'yea, there is a small deviation in share allocation but it seems to be working for other big projects like, kyber'.. or just not notice the issue at all which is probably the case for all the other projects up until the kyber situation.. But no we discovered the issue and continued to dig and dig until we came to the conclusion that the model is just not viable and we need something better. We have built this improved model and results are looking good. In the mean time we have not just been sitting back, while we have been testing we have been improving our algo and preparing for launch‼️ https://twitter.com/blackcubefi/status/1730305331077771514 Follow, Retweet, Like & Support❤️