
Nicolai Schneider
1.8K posts

Nicolai Schneider
@schneiderdotdev
dev; focus on digital identitification and iOS development; I also build apps on the side;






NEW: 🇩🇪🇨🇳 German Chancellor Merz says Germans need to work more in order to match China: “We are simply no longer productive enough. Each individual may say, “I already do quite a lot.” And that may be true. But when you return from China, ladies and gentlemen, you see things more clearly. With work-life balance and a four-day week, long-term prosperity in our country cannot be maintained. We will simply have to do a bit more.”

The timing is not ideal because every bozo is currently building some prediction bot and i've been keen to look into this field for a minute - my inspiration came from the book about Jim Simons ("The Man Who Solved the Market"). The idea of having an algorithm work 24/7 for you making emotionless trades to multiply your money sounded fascinating to me .. so i looked into it and openclaw couldn't have come at a better time. What most inspired me about Jim Simons is the fact that this man had no background in finance. The people he hired were mathematicians and together they simply analyzed data and tried to re-model it to predict future outcomes with impressive results (eventually they got to 80% anual returns). I may not have the advanced math skills a Jim Simons had but i live in a different time than when he started his trading company. A time full of new tools and easy access to information (he had a team digitalizing old market data when he first started). By. The. Way. I'm not going to act and say i wasn't intrigued by Polymarket's 15 minute btc bets, however, polymarket is banned in Germany so i had to move on from that thought rather quickly. Still, i thought the approach of just predicting whether a crypto coin would go up or down in the short term felt quite exciting and i decided to give it a shot. My idea was (is) to find a little workaround through Perps (basically a long/short bet an a coin without it ever ending - I decide when to close the position). I did some simple simulations (i let Opus 4.6 do it lol) and it appears as if 65% is the magic number of correct predictions to be profitable in the long run. My thought: i can do that or should at least try. So i started slightly clueless at first to predict BTC, Solana and Ethereum. My approach every 15 minutes: 1. If existent: evaluate the previous prediction 2. Gather current data (a snapshot from the current polymarket 15 minute bet, 5/10 EMA, RSI, Volume, News sentiment, Reddit sentiment) 3. Evaluate based on some given weights 4. Predict whether the coin goes UP or DOWN At first that went surprisingly well. I partially reached 70% accuracy over the course of 15-20 predictions (attached image). However, if such a simple approach would've made me profitable the entire market wouldn't really work and my system started crumbling and the number of false predictions increasing. None of the indicators appear to constantly be right and i need to dig deeper. Though that's ok because I started building a great workflow i can improve with ease. I mentioned that i use OpenClaw. There's a few helpful things i did. 1. The data retrieval and evaluation are all in dedicated python scripts and interchangeable 2. All predictions are logged as a json file including their decision parameters 3. Strategy changes are all documented as separate files 4. The 15-minute prediction is a cronjob and and doesn't cost me a cent (in LLM costs), however, if i want to make changes to the scripts i can just reach out to openclaw (Kimi K2.5) using telegram where ever i am and it will make the respective changes/fixes for me The combination of the above mentioned approaches have been incredibly helpful to set up my predictor, gather data and improve iteratively. I plan on sharing my journey as well as the ups and downs over time. Eventually, once i feel confident enough I will start trading with real money.




The timing is not ideal because every bozo is currently building some prediction bot and i've been keen to look into this field for a minute - my inspiration came from the book about Jim Simons ("The Man Who Solved the Market"). The idea of having an algorithm work 24/7 for you making emotionless trades to multiply your money sounded fascinating to me .. so i looked into it and openclaw couldn't have come at a better time. What most inspired me about Jim Simons is the fact that this man had no background in finance. The people he hired were mathematicians and together they simply analyzed data and tried to re-model it to predict future outcomes with impressive results (eventually they got to 80% anual returns). I may not have the advanced math skills a Jim Simons had but i live in a different time than when he started his trading company. A time full of new tools and easy access to information (he had a team digitalizing old market data when he first started). By. The. Way. I'm not going to act and say i wasn't intrigued by Polymarket's 15 minute btc bets, however, polymarket is banned in Germany so i had to move on from that thought rather quickly. Still, i thought the approach of just predicting whether a crypto coin would go up or down in the short term felt quite exciting and i decided to give it a shot. My idea was (is) to find a little workaround through Perps (basically a long/short bet an a coin without it ever ending - I decide when to close the position). I did some simple simulations (i let Opus 4.6 do it lol) and it appears as if 65% is the magic number of correct predictions to be profitable in the long run. My thought: i can do that or should at least try. So i started slightly clueless at first to predict BTC, Solana and Ethereum. My approach every 15 minutes: 1. If existent: evaluate the previous prediction 2. Gather current data (a snapshot from the current polymarket 15 minute bet, 5/10 EMA, RSI, Volume, News sentiment, Reddit sentiment) 3. Evaluate based on some given weights 4. Predict whether the coin goes UP or DOWN At first that went surprisingly well. I partially reached 70% accuracy over the course of 15-20 predictions (attached image). However, if such a simple approach would've made me profitable the entire market wouldn't really work and my system started crumbling and the number of false predictions increasing. None of the indicators appear to constantly be right and i need to dig deeper. Though that's ok because I started building a great workflow i can improve with ease. I mentioned that i use OpenClaw. There's a few helpful things i did. 1. The data retrieval and evaluation are all in dedicated python scripts and interchangeable 2. All predictions are logged as a json file including their decision parameters 3. Strategy changes are all documented as separate files 4. The 15-minute prediction is a cronjob and and doesn't cost me a cent (in LLM costs), however, if i want to make changes to the scripts i can just reach out to openclaw (Kimi K2.5) using telegram where ever i am and it will make the respective changes/fixes for me The combination of the above mentioned approaches have been incredibly helpful to set up my predictor, gather data and improve iteratively. I plan on sharing my journey as well as the ups and downs over time. Eventually, once i feel confident enough I will start trading with real money.





Software engineering makes up ~50% of agentic tool calls on our API, but we see emerging use in other industries. As the frontier of risk and autonomy expands, post-deployment monitoring becomes essential. We encourage other model developers to extend this research.




















