Pricing algorithms
How a buy price and a sell price are derived from Steam price history, and which filters can veto a purchase.
Every price the bot uses comes from one analysis run over Steam's public price history for that item. The run produces a buy price, a sell price, and a verdict on whether the item may be bought.
Two sides of every price
Steam charges the buyer; the seller receives less. xBot tracks both:
| Name | Meaning |
|---|---|
| Steam price | What a buyer pays — the number shown on the Steam page. |
| Sell price | What the seller receives after Steam's fee. |
| Buy price | What the bot offers in a buy order. |
The conversion uses a fixed multiplier of 1.15:
steam price = sell price × 1.15
sell price = steam price ÷ 1.15The analysis window
- Steam's price history is read in the market timezone (America/Phoenix).
- Points are bucketed: period = the last Analysis period days ending yesterday, plus week (7 days), month (30 days) and today. Today's trades never enter the pricing window — only the trend filters read them.
- The Corridor filter drops points further than
corridor %from the bucket average. With0nothing is dropped. - If the period bucket is empty after filtering — or Steam returned fewer than four history points — the analysis is skipped: the item keeps its current prices and no error is raised.
Algorithms
All five run on the same window; the group picks one for the buy side and one for the sell side independently. "Daily high/low" means the highest or lowest price a calendar day traded at.
| Algorithm | Base price | Anchored to |
|---|---|---|
| Maximal | median of the daily highs | sell side |
| Safe | the lowest daily high | sell side |
| Average | mean of every trade in the window | sell side |
| Low | median of the daily lows | sell side |
| Buy | median of the daily lows | buy side |
Sell-anchored algorithms take their base price as the Steam price, strip the fee to get the sell price, then subtract the markup to get the buy price:
steam price = base
sell price = base ÷ 1.15
buy price = sell price − markup % of sell priceBuy goes the other way — the base price is the buy price, and the sell price is the price that yields the configured margin:
buy price = base
sell price = buy price ÷ (1 − markup / 100)
steam price = sell price × 1.15Low vs Buy
Both read the median of the daily lows. They differ in what that number becomes: Low treats it as the price buyers pay (so your buy order lands a markup below it), Buy treats it as your own bid (so your listing lands a markup above it). Buy is the more aggressive of the two.
The Average algorithm reads the history before the corridor filter, so it is the only mode that ignores Corridor, %.
Scaling
The chosen prices are then scaled by their own shares:
buy price = buy algorithm price × (Buy algorithm, % ÷ 100)
sell price = sell algorithm price × (Sell algorithm, % ÷ 100)The Steam price is re-derived from the scaled sell price. A share of 0 is treated as 100.
Outbidding
With Outbid the highest buy order enabled, a calculated buy price above the highest competing buy order is pulled down to that order plus one cent, never above the calculation. This account's own unfilled orders are excluded. The option also forces an order-book request on every analysis.
With Undercut the cheapest listing enabled,
a fresh book at placement targets the cheapest competing listing minus one cent, excluding
this account's own active listings. Seller proceeds after fees must satisfy both positive
profit limits: price_buy × percent / 100 (rounded up to a minor unit) and the fixed amount
in account currency. The floor is purchase price plus the larger minimum, not the calculated
sell price. Zero disables an individual constraint. This mode does not cancel existing listings.
Filters
Filters decide whether the item may be bought. They never change a price, and they never stop selling.
| Filter | Passes when |
|---|---|
| Sales per day | window volume ÷ window days (rounded down) ≥ the configured minimum |
| Weekly trend | today's average stays within the rise/drop limits around the 7-day average |
| Monthly trend | today's average stays within the rise/drop limits around the 30-day average |
| Daily trend | always — the comparison is against today's own average |
| Order book coefficient | quantity queued at or above your buy price ≤ daily sales × the coefficient |
| Position in listing | your Steam sell price lands no deeper than the configured position in the listing queue |
A filter set to 0 is disabled. One failing filter is enough to block the purchase; the item is
still listed and sold as usual.
Choosing a mode
| Goal | Starting point | Check |
|---|---|---|
| Conservative first run | Safe for both sides | Prices on several items before enabling the group |
| Anchor to the lower part of the range | Low | The resulting margin after Steam's fee |
| Bid from daily lows | Buy | This is more aggressive: the base point becomes the bid itself |
| Use the mean of all trades | Average | This mode ignores the corridor filter |
| Anchor to typical daily highs | Maximal | Outliers and liquidity over the chosen period |
An algorithm name does not guarantee a profit. Start with a small group, compare calculated prices with the market, and only then enable buying and selling.
Worked example
A group with Safe on both sides, markup 3 %, shares at 100 %, over a 5-day window whose daily
highs are 2.30, 2.20, 2.00, 2.45, 2.10:
base (lowest daily high) = 2.00 → Steam price 2.00
sell price = 2.00 / 1.15 = 1.74
buy price = 1.74 − 3 % = 1.69The bot bids 1.69, lists at a Steam price of 2.00, and nets 1.74 per sale before the market moves.