Partial solution to the Stable routing bonus quiz:
I've mapped stable-to-stable across DEXen. Here're those mappings:
You see from the raw data (below), there are 66 edges to this graph. SOME of which have a gain-multiplier (marked in green)!
Partial solution, part II:
- We use @neo4j Aura graph database.
- We upload using their graphical uploading tool, which gets easier than "LOAD CSV" once you get the hang of it.
- Sample Stable Coin graph
- Dijkstra shortest path $DAI.e -> $USDT.e
But there's a problem...
The problem here is Dijkstra solves for SHORTEST path, but the LOWER the multiplier, the MORE disadvantageous the trade.
I must rework my data set that has an inverted multiplier, but some multipliers, as you saw, are ABOVE 1.0, i.e.: an ADVANTAGEOUS swap.
Some thought required.
Okay, redesigned with an efficacy-score. DONE! ✅
NOT DONE! 😤
The problem with pathing algorithms for this particular domain is that the multipliers are not path-lengths, but, as their name denotes: multipliers.
I think I may have to tune an algorithm then patten(tm)(r)(c) it.
BUT IN THE INTERIM!
I CAN query the graph and ask:
"Are there paths that pay ME to swap?" or, in cypher:
match p =(c)-[r]->(c1)where r.efficacy < 1return p
And there are:
How about we go one step further?
"Is there a multiple hope currency-swap that is efficacious?" or, in cypher:
match p =(c)-[r]->(c1)-[r1]->()where r.efficacy < 1 and r1.efficacy < 1return p
Oh, well.
No comments:
Post a Comment