Content
@
https://opensea.io/collection/dev-21
0 reply
0 recast
2 reactions
Makushi π§
@makushi
A month ago, I shared some theoretical work on how vector embeddings could be used to build market predictions. Today, let's dive into some of the integration details behind the analysis, and give a peek at the results. πβ¬οΈ
1 reply
0 recast
0 reaction
Makushi π§
@makushi
π’ Performance Evaluation Using a daily CRON job, the system (letβs call it Viktor from now on) starts by evaluating the previous dayβs predictions. It fetches the current prices of all selected tokens and records them to track daily performance easily.
1 reply
0 recast
0 reaction
Makushi π§
@makushi
π΅ Token Discovery Next, Viktor uses the Mobula APIs to query all listed tokens (~27,000), then filters them based on: - Minimum market cap - Minimum liquidity - Exclusion of stablecoins - Compatibility with supported chains (enabling auto-trading via Uniswapβs SwapRouter)
1 reply
0 recast
0 reaction
Makushi π§
@makushi
This initial filter leaves us with ~300 tokens on average. These are scored using volatility, volume, liquidity, market cap, and social metrics, then sorted to keep only the most active and liquid ones.
1 reply
0 recast
0 reaction
Makushi π§
@makushi
Finally, Viktor performs a check for the presence of ETH or USDC pools on each tokenβs chain using RPC calls to the Uniswap Factory contract of the tokenβs chain. After this, the token discovery phase is complete and ready for vector analysis.
1 reply
0 recast
0 reaction
Makushi π§
@makushi
π‘ Analysis and cosine similarity search For each token, Viktor fetches the last 10 days of OHLCV data (Open, High, Low, Close, Volume, which is the underlying data of market charts). This window is then normalized and converted to natural language and embedded as a vector and compared, via Supabase pgvector, against ~56,000 (at time of writing) training windows stored in memory.
1 reply
0 recast
0 reaction
Makushi π§
@makushi
This lets us compute weighted averages for each outcome group, estimate potential returns, and assign a confidence score to the forecast. Tokens are then ranked based on their potential for positive price movement in the days ahead. π
1 reply
0 recast
0 reaction
Makushi π§
@makushi
βͺ Results and conclusion All daily results are saved on Supabase and can be monitored on an interface (https://viktor-monitor.wakushi.com) so I can track the evolution of performances as Iβm improving the system and tweaking some of its rules. This first version of Viktor was launched after achieving 55-60% of good predictions by comparing past windows analysis against its training vector embedded data on thousand of days.
1 reply
0 recast
0 reaction
