All Games8 min read

How the Flip Finder Works

The Flip Finder is not a black box. This guide explains exactly how items are scored, why different games use different models, where the underlying data comes from, and how to interpret — and sanity-check — the scores you see.

What the Flip Finder Does

Each supported game has thousands of tradeable items. Manually scanning every item for a profitable trade opportunity would take hours. The Flip Finder solves this by automatically calculating a composite score for every item in the database, then surfacing the top-ranked items in a sorted, filterable table.

The scores are recomputed from the latest price snapshot data after each price refresh cycle. For GW2 and FFXIV this means scores are recalculated every 30 minutes. For WoW Classic, scores refresh roughly every 65 minutes (after the 60-minute price pull completes). The “last updated” timestamp in the Flip Finder header shows when the current set of scores was computed.

The model differs significantly between GW2/FFXIV (which have live price data that supports spread-based analysis) and WoW Classic (which has a one-sided AH that requires a different approach). Path of Exile does not use the Flip Finder at all — it uses the Investment Analyzer instead. Each model is explained below.

The GW2 Spread-Based Score

GW2's two-sided market (buy orders + sell listings) allows a spread-based scoring model. The spread is the gap between the highest buy order and the lowest sell listing. A flip exploits this gap: place a buy order, wait for it to fill, then post a sell listing at the ask.

The GW2 score is computed from four components:

ComponentWeightWhat it measures
Spread as % of sell price40%Primary signal. How wide is the gap relative to the item's price level? Wider = more arbitrage room after fees.
Buy price vs. 30-day average30%Is the buy side currently depressed vs. its recent norm? Adds confidence that the spread is a temporary opportunity, not a permanent structural feature.
Transaction velocity20%How often does this item trade? A wide spread on a market that rarely transacts means your order may never fill.
Volatility penalty10%Items with erratic price history are penalised. Volatile markets make the expected spread unreliable.

The post-fee profit shown in the table already accounts for GW2's 15% total fee (5% listing + 10% exchange). You do not need to calculate fees separately — what you see in the Profit column is what you would actually receive if the flip executed at the current prices.

The WoW and FFXIV Temporal Score

WoW Classic and FFXIV both have one-sided markets (sell listings only, no standing buy orders). Without a live buy/sell spread to exploit, the spread model does not apply. Instead, AuricDB uses a temporal model: compare the current price to the 30-day historical average and score how underpriced the item currently is.

The core bet in temporal scoring is mean reversion: items that are currently cheap relative to their historical norm tend to recover toward that norm. You buy at the depressed price and relist at (or near) the historical average.

ComponentWeightWhat it measures
Current price vs. 30-day averagePrimaryHow far below the historical average is today's lowest listing? A large gap is the main signal that a mean-reversion opportunity exists.
Listing count and velocitySecondaryIs the market liquid enough to sell into? Very low listing counts mean a thin market where your relist may take days or weeks to fill.
Volatility penaltyTertiarySame as GW2: erratic price history reduces confidence that the 30-day average is a meaningful recovery target.

An important difference from the GW2 model: temporal scoring is inherently a multi-day hold strategy. On GW2 you can place a buy order and have it fill within minutes on a liquid item. On WoW and FFXIV you must actively buy the item and then wait for your listing to sell — which could take hours or days. The risk profile is different: you are tying up your currency for a longer period, and the price could change further before your listing fills.

The temporal score requires at least 3 historical data points for an item before it can be computed. Items without sufficient history are not included in flip results even if their current price looks attractive.

Why PoE Uses the Investment Analyzer Instead

Path of Exile's player-to-player economy has no standing buy orders and no automated exchange. There is no “buy price” in the same sense — you message a seller and they choose whether to trade. This means there is no live spread to exploit and the temporal model also breaks down (you cannot buy an item at the lowest listing price and immediately relist at the 30-day average — the market does not work that way).

PoE's economy is better understood through league lifecycle patterns: which items tend to appreciate at which phase of the league, and which signals indicate that a particular item is currently in its buy zone. This is what the Investment Analyzer measures — see the PoE League Investment Guide for the full explanation.

Data Sources and Refresh Cadences

Guild Wars 2

Source

Official GW2 Commerce API (ArenaNet)

Price refresh

Every 5 minutes

GW2 provides the highest-frequency data of any supported game. Scores are recomputed from fresh 5-minute snapshots every 30 minutes.

WoW Classic

Source

Blizzard Auction House API

Price refresh

Every 60 minutes (per realm)

Blizzard's AH API rate limits enforce a maximum of roughly one full dump per hour per realm. Scores are computed after each hourly pull completes.

Final Fantasy XIV

Source

Universalis (community aggregator)

Price refresh

Every 30 minutes (per world)

Universalis aggregates real-time Market Board data from players running the Universalis plugin. AuricDB queries Universalis every 30 minutes per world.

Path of Exile

Source

poe.ninja

Price refresh

Every 30 minutes

poe.ninja aggregates listed prices from player shops. Investment Analyzer scores recompute after each pull.

Interpreting the Score: What Is a “Good” Number?

The score is a relative ranking within a game at a point in time, not an absolute measure of expected return. A score of 80 means this item is in the top tier of all currently tracked items in this game — not that you will make an 80% profit. The actual profit depends on the item's absolute price level and your trading volume.

This is why the Minimum Profit filter exists. An item scoring 90 with a 2-copper post-fee profit is less valuable than an item scoring 65 with a 50-silver profit. Use the score to find items with strong arbitrage or mean-reversion signals, and use the profit figure to filter down to items where the absolute return justifies the effort.

Colour coding in the Flip Finder table provides a quick visual orientation:

Score ≥ 60 — strong signal, multiple components aligned
Score 30–59 — moderate signal, worth investigating
Score < 30 — weak signal, low priority

What the Score Does Not Tell You

The Flip Finder is a filter, not a decision engine. Several important factors are outside its scope:

  • In-game execution speed. A score computed 25 minutes ago may reflect a spread that has already been arbitraged away by other players. High-score items on liquid markets attract competition.
  • Crafting routes. An item with a wide spread may be craftable by players for less than the buy order price, meaning buy orders will never fill. The score has no knowledge of crafting recipes or material costs.
  • External market events. A patch announcement, a streamer promoting a specific build, or a server-side event can invalidate a score instantly. The data is a snapshot; the world moves on.
  • Stale sell listings. A sell listing posted months ago and never cancelled creates a false impression of the ask side of the market. The spread score will see a wide spread; the reality is that nobody is actually selling at that price actively.

Always look at the item's price history chart before committing. Thirty seconds of chart review — checking that the spread is real, the volume is present, and the trend is not against you — prevents the most common flip mistakes.

Related guides