FFXIV Arbitrage Engine

What is the FFXIV Arbitrage Engine?

This engine uses Universalis to scan across multiple worlds and datacenters to find underpriced items you can buy low and sell high - what we call flips or arbitrage targets.

It works by comparing current listings against recent sales history. If an item is listed well below the average sale price on a high-velocity world, it shows up as a candidate. You then buy it and bring it back to your home world to sell for a profit.

The static pages (Balmung, Malboro, Mateus, Seraph) show pre-computed scan results updated hourly. The Scanner lets you run a fresh live scan with custom filters and parameters.

How It Works

Fetch marketable items: We pull the list of items that actually trade on the market board.
Batch-fetch listings and history: For each world, we grab current listings and recent sales data from Universalis.
Compute candidates: For every item, we calculate the gross profit (sell price minus buy price plus fees), velocity (how fast it sells), and confidence (how recently it sold at that price).
Score and rank: Each candidate receives a composite score that balances profit, turnover, and certainty. The highest-scoring items rise to the top.
Display results: Results are organized into tiers (Gillionaire, High, Mid, Low) so you can pick the risk and reward level that fits your playstyle.

Metrics: Velocity & Score

Velocity (Vel/d)

Velocity is the average number of units sold per day on the sell world. It tells you how quickly you can move inventory.

Higher velocity means faster gil turnover and less risk of sitting on stock. We use a square-root dampening function so extreme outliers don't completely dominate the score.

Score

The score is a single number that balances profit, velocity, and confidence into one comparable metric. The formula is:

score = ( gross × vel × conf × supplyFactor ) / 10

gross = sqrt( min( profit_per_unit, 1,000,000 ) )
Profit per unit is capped at 1M gil and square-rooted to dampen outliers. Without the cap, something like an Earth Shard listed for 999,999,999 gil would break the equation and push every real item off the board.

vel = sqrt( max( 1, daily_sales_velocity ) )
Velocity gets the same square-root treatment so a 600/day item doesn't score 10x a 60/day item - the difference matters, but it shouldn't be all-consuming.

conf = max( 0.05, last_sale_price / current_home_price )
Confidence is the ratio of the most recent sale price to the current listing price, clamped to a floor of 0.05. If we have no recent sale data, we still give the item a small non-zero score rather than discarding it entirely.

supplyFactor = linear ramp from 1.0 (fewer than 3 listings) down to 0.05 (20 or more listings)
When a world is flooded with copies of an item, your chance of selling drops. This factor penalizes oversupplied markets.

/ 10: Brings the final number into a readable range.
The raw product of these factors can still span a wide range. Dividing by 10 keeps scores in a human-friendly scale without changing relative rankings.

Why We Rank This Way

Pure profit sorting would surface massive, slow-moving items that tie up capital for weeks. Pure velocity sorting would show cheap consumables that return pennies. The score is designed to find the sweet spot: items that move quickly, have a reasonable margin, and aren't drowning in competitor listings.