We are happy to announce the final rankings for Cryzen Battle of the Bots #1!
- Brenton — at $95.78 USDT (4.83% over buy and hold)
- Shuvro — at $95.36 USDT (4.42% over buy and hold)
- Terje — at $94.79 USDT (3.85% over buy and hold)
- Yash — at $90.60 USDT
- Jason — at $87.78 USDT
Congrats to Brenton our final winner! And recipient of $1k USD and 50k XEN token for first place.
2nd and 3rd place goes to: Terje and Jason, to win 30k and 20k XEN tokens respectively (Shuvro and Yash do not qualify for 2nd and 3rd place award sums as they are Cryzen affiliated).
All bots started with $100 USDT and the ending balances are given above. As mentioned in a previous post, all bots have a current balance lower than the starting balance at the beginning of the competition. Nonetheless, 3 out of 5 bots have outperformed buying and holding etc/usdt, which has declined in value from the start of the competition by 9.05%.
This indicates the need for bots capable of shorting as described in our previous competition update here.
Winners decided based on backtesting
Final winners were decided based on backtesting the 5 valid submissions above, from April 14th to May 12th. The bots were initially run live — On May 8th, however, Binance deleted all api keys for all of their users (api keys are needed for Cryzen bots to place live trades on behalf of our users), resulting in all of the competition bots turning off.
As a result, we have decided to pick the winners based on backtesting from April 14th to May 12th instead (the duration for which the bots were to run live). Rankings on live runs and backtests were not different on the weekly checks we have done from April 14th to May 8th, which is great news for the Cryzen backtesting system (in fact backtests on average were within .1% of live runs in terms of profits reported).
The Winning Bot
We will interview Brenton, the winner of this competition, and have a follow up article detailing his bot. Stay tuned, and happy trading!
What we learned?
During the development of bots for this competition, it became very clear that each bot, with a few set of parameters, can perform in vastly different ways based on the chosen values for these parameters.
With just a few parameters for a bot, the possibilities quickly blow up. For example, the bot above has 4 main parameters, if we are interested in 10 values per parameter, there are 10,000 different possibilities.
It is tedious and not possible to fully explore these combinations manually. One much needed feature is automated strategy optimization and exploration.
To this end, Cryzen is releasing a feature next week to do just this — push one button and get a 3D plot of the profitability of a strategy across a multitude of the combinations of its parameters.