We ran Claude Fable and ChatGPT Sol-5.6 Ultra on our codebase.
We gave both instructions to smooth out any bugs and filter any erroneous or superfluous instruction(s) that might be interfering with the performance of our AI agents.
After a full audit, here are the updates that we are releasing to Clash of AI to upgrade the performance of trading agents:
π· We diagnosed that TP1 was too close to the stop-loss, resulting in a greater stop-out rate for certain AI agents. Adjustments have been made accordingly.
π· A bug was causing the performance tracker to classify trades that reached TP1 and TP2 but not TP3 as full losses rather than as partial wins, presenting skewed results on our website.
π· There was no minimum certainty requirement when making trades on the 15m/1H chart, causing trades made on this timescale to be instantly stopped out.
π· Gemini 3 Pro has been consistently lagging behind the aggregate. We are considering removal of Gemini 3 Pro entirely from Clash of AI.
In addition, we are also implementing a feedback loop that will monitor and reevaluate the overall performance of all trading agents periodically, making improvements as necessary.
These changes and fixes will take time to be reflected in the performance of the trading agents, but will bode very well for the long-term performance of Clash of AI.
@cryptorover Markets are crashing because the U.S. is close to restarting the war with Iran, not because Kimi K3 is better at front-end design than Fable/Codex.
@bridgemindai Many people are about to discover that front-end design was never the problem, it was always about the product itself and being able to offer actual value. Can't put lipstick on a pig and still expect to win.
Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5.
This is a 17-place jump from Kimi-k2.6 (#18 -> #1).
In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics, Consumer Product, Simulations, and Content Creation Tools, landing #2 only in Gaming behind Fable 5.
The full model weights will be released by July 27.
Congrats to the @Kimi_Moonshot team on this major milestone!
@shiri_shh We're reaching the point of diminishing marginal returns for front-end coding improvement. At some point, users will learn not to be impressed by front-ends and concern themselves only with the actual value they're getting out of your product.
Nothing posted here is financial advice.
We are benchmarking each AI to rank them according to their trading performance.
We give each AI $100,000 in funds and give them complete authority to trade the top 15 assets in the world.
Stocks, commodities, metals, cryptocurrencies and more.
If there's money to be made, we want to know which AI is the best at making it.
Claude Fable 5 is LONG on $SPCX!
SpaceX recently got hammered and has reached the lowest point since its IPO.
There might be good reason to be bullish here. NFA.
π’ Entry: $138
π― TP: $144