Undercutting, price walls and where to place your offer
You know the spread. Now you have to choose a number inside it, and that choice decides whether you fill in forty seconds or never. The whole skill is guessing how many people are standing at each price ahead of you.
Two rules from lesson 1 do all the work here. A trade happens when a buy offer meets or exceeds a sell offer. Where several offers share a price, older ones are loosely prioritised over newer ones. Put those together and every price level in the game is a queue, and joining a price means joining the back of it.
Reading a queue you cannot see
Nothing in the game tells you how many units are stacked at a price. You have to infer it, and there are only three signals available.
- Daily volume. An item trading 49 million times a day has queues at every price near the middle. An item trading 400 times a day has essentially nobody standing anywhere. Volume is on every item page on this site and is the single best proxy for queue depth.
- How your own offer behaves. An offer that fills in a steady trickle is at the front of a moving queue. An offer that fills nothing for ten minutes and then twenty units at once is behind something that just got cleared.
- Whether the price moves while you wait. If trading is clearly happening and the price is not moving, there is size at that level absorbing it.
That is genuinely all you get. Anyone claiming to show you Grand Exchange order book depth is showing you a model, not data.
The one gp question
You measured the highest bid at 771. Do you join it, or bid 772?
The gp itself is never the issue. The issue is what the gp buys, and that depends entirely on how long the queue at 771 is.
When the extra gp is the cheapest thing you buy all session
On a heavily traded item, matching the current best price puts you behind hundreds of other offers. Your fill rate is whatever leaks past them. One gp higher creates a new, empty level, and every unit sold into the market hits you first until someone else does the same thing.
On Magic logs, trading over five and a half million units a day, the difference between the front and the back of a queue is minutes against hours. The buy limit is 12,000 per four hours, so the whole trade is a race against a clock. Spending one gp per unit to be first is 12,000 gp across a full limit, which is around a tenth of the profit and easily worth it if it turns one filled window into two.
When the extra gp is wasted
On a thin item there is no queue to jump. If Dragon claws trade about eleven hundred times a day across the entire game, that is roughly one trade every eighty seconds worldwide, and the number of resting offers at any single price is small. Bidding one gp over on a 38 million gp item is not aggression, it is noise. The ladder there moves in tens of thousands.
On expensive, thin items the meaningful increments are percentages, not coins. If you want priority you have to move the price somewhere a seller will notice, and that is a completely different decision with a completely different cost.
The rule of thumb. If one gp is a rounding error against the item's price, it is also a rounding error against the queue and buys you nothing. If one gp is a visible fraction of the spread, it buys you position. Cheap high volume goods sit in the second case, expensive rare goods in the first.
The free undercut at multiples of 50
Here is a genuine quirk of the tax rules, documented on the wiki, that almost nobody uses.
The 2% tax rounds down to the nearest whole coin. At a sale price of 100 gp the tax is 2 gp, so the seller keeps 98. At a sale price of 99 gp the tax is floor(1.98), which is 1 gp, so the seller also keeps 98. The two prices pay the seller exactly the same amount.
This generalises to every exact multiple of 50. Undercutting a sell offer that sits at 50, 100, 150, 200 and so on by a single gp costs the seller nothing at all, while putting them ahead of everyone parked on the round number. Since players love round numbers, sell offers pile up on them, and stepping one coin below a round number is often the largest queue jump available anywhere on the Grand Exchange for a price of zero.
It only matters at prices low enough for single coins to exist meaningfully, so think cheap consumables rather than gear. It is still free money for a single keystroke.
Price walls
A wall is a very large resting offer sitting at one price. You never see it directly. You detect it by its effects.
On the sell side, a wall means the price refuses to rise. You place a buy above it, fill instantly at the wall's price, place more, fill again, and the guide value does not move. Every buyer in the game is being served by the same stack. On the buy side, a wall means the price refuses to fall: sells keep getting absorbed at one number and the item stops dropping.
Walls come from three very different sources, and telling them apart is most of the value in noticing one.
- Genuine supply or demand. A popular consumable simply has thousands of players on each side. This kind of wall is stable and boring, and it is why the price of a common rune barely moves for weeks.
- One large holder liquidating. Someone with a huge stack is feeding it out at a fixed price. This wall is finite. When it clears, the price often moves quickly, because it was the only thing holding the level.
- A deliberate placement. Someone wants the price to look supported. That wall disappears the moment it has done its job, and it is the tell described in lesson 7.
A wall that vanishes without being traded through is the dangerous one. If a level has been absorbing everything for an hour and then simply is not there any more, the support you were relying on was never real.
Offers that never fill
An offer that sits untouched for hours is almost always one of five things.
- You are behind a queue. Common on liquid items at a matched price. Fix: reprice by one step and you jump the whole line.
- You are behind a wall. Your offer is at the same price as thousands of units that were there first. Fix: step past the wall or accept the wait.
- The market moved away. Your buy at 771 was fair when you placed it and the item is now bid 790. Nothing will ever come back to you. Fix: cancel and reprice, do not wait for the market to return.
- There is no other side. Some items genuinely have no resting bids. Your margin check would have caught this, because the sell leg would not have filled either.
- You hit the buy limit. The offer looks alive but has stopped buying because you have taken your four hour allowance for that item. This one is invisible if you are not tracking it.
That last one is worth solving rather than remembering. Track each flip as you place it and My Flips holds your open positions in one view, so an offer that has quietly stopped buying is visible as a position rather than a mystery. With the RuneLite plugin linked to your account, your Grand Exchange fills arrive there without you typing anything, which is the only version of this that stays accurate across a long session with eight slots turning over.
Three of the other four are diagnosable before you place at all, and the Margins board is where. Trades/h tells you whether there is a queue to be behind in the first place, since a busy item has one and a thin item does not. Last buy and Last sell tell you whether the other side exists right now: a row where the sell side last traded forty minutes ago is the "there is no other side" case, visible in advance instead of after an hour of watching an offer do nothing. Filter on those three before you decide whether to match, step or wait, and most of the placements that were never going to fill are gone before you make them.
Cancelling is cheap. There is no penalty for aborting an offer and collecting what filled, and a slot doing nothing is worse than a slot doing something small. Eight slots is the real budget in a flipping session.
Two placements, same session
Magic logs, bid 771, ask 795, limit 12,000, volume around 5.6M a day.
Tax on a 795 sale = floor(0.02 × 795) = 15 gp. Net per log at 771 in, 795 out = 9 gp.
Full limit at 9 gp = 108,000 gp per window on 9.25M of capital.
Bidding 772 instead drops you to 8 gp, so 96,000 gp. You gave up 12,000 gp of profit to be first in line.
Worth it if it turns a half filled window into a full one. Not worth it if you were filling anyway.
Dragon claws, around 38M, limit 8, volume around 1,120 a day.
One gp over on a 38M item is 0.0000026% of the price. There is no queue for it to jump.
To gain real priority here you would move the price by a meaningful fraction, perhaps 50,000 to 100,000 gp, and that comes straight out of a spread you can count on one hand per window.
The correct move on a thin, expensive item is usually to place the price you want and wait, and to size so that waiting is comfortable.
Placement rules worth keeping
- Match the best price only when you are happy to wait behind everyone already there.
- Step one increment when volume is high and the increment is small relative to the spread.
- Never step when the increment is invisible against the item's price.
- Undercut a round number by one coin whenever the item is cheap enough for that to be a real price.
- Treat an unmoving price with heavy trading as a wall, and decide whether you want to be behind it or past it.
- Cancel early. An offer that has not moved in an hour has told you something, and holding it does not change the answer.
Placement decides how you get in. The next lesson is about how you get out when the plan has already gone wrong, and how much that exit is really costing you.
Item pages used above
Live prices, volume and buy limits for the items in this lesson: Magic logs, Dragon claws and Nature rune.
GE Uncut is an unofficial, fan-made tool, not affiliated with or endorsed by Jagex Limited. RuneScape and Old School RuneScape are trademarks of Jagex Limited; in-game content and item images are the property of Jagex Limited, used for reference only. Price data is from the OSRS Wiki real-time prices API. See data attribution.