Following my previous work, I turned my attention to bananas. Their price movements were unstable, making it impossible to trade around a fixed price like I did with pearls. So I decided to first examine the order book dynamics for bananas.
After analyzing the bid-ask spread each round, I made an interesting discovery: the spread size varied significantly. Most of the time it was above 4, butabout201. This was particularly intriguing because when the market spread narrowed to $1, it suggested one side (either bid or ask) was likely mispriced, creating a potentially profitable opportunity.
I then investigated these 20% cases with tight spreads. What I found was that these situations typically occurred suddenly – for example, when the spread had been consistently wide but suddenly tightened to 1, onlytoquicklyreturnto4 shortly after.
Looking at a specific example, I noticed a pattern: at one moment the best bid was 4896 (20 sets) and best ask was 4900 (20 sets). Then someone inserted a small ask order at 4897 (3 sets) between them, causing the spread to suddenly shrink to $1. This small order would quickly get filled, and the spread would return to its previous width.
I realized these small orders represented the trading opportunities I wanted to capture. The challenge was how to systematically identify such situations. After much thought, I developed a solution:
1. Calculate a volume-weighted average bid price (to reduce noise from small orders)
2. Similarly calculate a weighted average ask price
3. Average these two to get a weighted mid-price
Using the earlier example:
Weighted mid-price = [4896 + (4897×3 + 4900×20)/23]/2 = 4897.8
Now when that anomalous 4897 order appeared (below my weighted mid-price), I would aggressively buy. I adapted my pearl Strategy v2, using this weighted mid-price as bananas’ dynamic fair price, creating Strategy v3. The results were excellent, generating over 2,000 in profits from banana trades.
