Commission tiers: does the lower rate apply to every lot, or only to the lots above the threshold?

[Fairman]

Commission tiers: does the lower rate apply to every lot, or only to the lots above the threshold?

A lot of raw spread accounts list commission by monthly volume. Take these illustrative tiers, not from any particular firm: $7 per lot round turn up to 50 lots, $6 from 50 to 200 lots, $5 from 200 lots up. The page rarely says how the tiers are applied, and the answer changes the bill. Say you trade 120 lots in a month.

Marginal, like income tax bands: the first 50 lots at $7 is $350, the next 70 at $6 is $420, total $770. That's an average of $6.42 a lot.
Retroactive: once you pass 50 lots, all 120 are charged at $6, total $720.
Based on last month's volume: if last month was under 50 lots, all 120 are charged at $7, total $840.

Same trading, three bills, and a $120 gap between the highest and the lowest. On a $10 pip, each dollar of commission per lot is 0.1 pip, so for someone whose targets are 4 or 5 pips, the gap between schedules is a real part of the edge.

One more way to get it wrong: if the page quotes commission per side, $7 is $14 round turn, and reading it the other way round halves or doubles the bill before any tier question comes up.

Questions I'd put to the support desk before picking an account: are the tiers by calendar month or a rolling 30 days, are lots counted per side or per round turn, and is the tier worked out per account or across all accounts under one client?

[FXS HongKong Scalper]

The difference between marginal and retroactive tiers can be large. One more detail is how volume is counted: some brokers count only one side of the trade, others count both open and close. That can halve or double the volume needed to reach a tier. It is worth asking the broker to explain with a worked example, as you did here. For traders close to a threshold, knowing whether the lower rate applies to all lots can change whether it is worth trading slightly more in a month.

[Fairman]

If the tiers are retroactive, there's a cliff. Using the same illustrative schedule, 199 lots all at $6 costs $1,194, while 200 lots all at $5 costs $1,000. One extra lot lowers the whole bill by $194. A trader sitting at 195 lots on the 28th has a financial reason to add trades just to cross the line, which is a bad reason to trade. Knowing a cliff exists is useful mainly so you can ignore it deliberately. If the schedule is marginal, there's no cliff, the cost per extra lot just steps down. It's another reason to find out which scheme applies before you plan around the tiers.

[Amanhasnoname]

I asked exactly this question to my broker's support last year and it took three emails to get a clear answer. Their page listed tiers by monthly volume, and the first reply just repeated the table. Only when I gave them a worked example with a specific lot count did they confirm that the tiers were based on the previous month's volume, so a big month only got the cheaper rate the following month.

That changed how I thought about it. My volume swings a lot between quiet summer months and busy ones, so the discount often applied in a month when I traded little and missed the month I traded most. Over a year, the effective rate was much closer to the top tier than I had assumed.

The per side versus round turn issue tripped me up once as well. I compared two brokers and thought one was half the price of the other, until I realised one quoted per side and the other round turn. They were actually the same cost. Now I convert everything to round turn per standard lot before comparing anything.

[Fairman]

Here's what 0.2 pip does to a scalp. Take a 4 pip stop with a 5 pip target. Break-even win rate is (stop plus cost) divided by (stop plus target). With a total cost of 0.9 pip that's 4.9 divided by 9, 54.4%. Cut the cost to 0.7 pip and it's 4.7 divided by 9, 52.2%. Two percentage points of win rate for 0.2 pip, which is $2 per lot round turn. A trader who wins 53% of these trades loses money at 0.9 pip and makes money at 0.7. Edges at this scale are decided by details of the fee schedule more than by the chart.

[Fairman]

A concrete scalper: 5 trades a day, 1 lot each, 21 trading days, so 105 lots. On the illustrative schedule, marginal pricing gives 50 lots at $7 plus 55 at $6, which is $680. Retroactive gives 105 at $6, $630. Previous month's tier gives 105 at $7, $735. The spread between highest and lowest is $105, about 14% of the top bill. In a month where the strategy nets $600 before commission, that gap is 17% of the result, so the fee schedule can matter as much as a strategy tweak.

[Fairman]

A list for the chat window with support, so the answers are in writing. Which tier basis applies: marginal, retroactive or previous month? Does the month run by calendar or by rolling 30 days? Are lots counted per side or per round turn? Does a partial close count as one trade or two? Do multiple accounts under one client pool their volume? Does an introducing broker or cashback arrangement change the tier? Keep the transcript. Terms sometimes change quietly, and a saved chat with a date on it gives you something to point at if the first invoice doesn't match what you were told.

[Shadow Trader]

Count the number of price updates per second on a few pairs during the London session. EURUSD might refresh ten or twenty times per second. A cross like EURCHF may manage three or four. An exotic might update once every second or two.

That rate is a decent proxy for how many participants are quoting and how actively. It isn't perfect, since aggregation and filtering affect it, but it's far better than tick volume as a measure of liquidity because it describes the pace of the quoting side of the market.

A few uses stand out. A falling update rate with a stable spread suggests providers are stepping away quietly. A rising rate with a widening spread suggests nervous repricing. A pair whose update rate is too low for a strategy's holding time is simply not suitable, because the price a scalper sees may be stale.

Counting updates takes a few lines of code in most platforms. Logging the figure by time of day over a couple of weeks produces a liquidity profile for each pair that is more useful than any general guideline about sessions.

Has update frequency ever been used as a filter, or is it only looked at after something goes wrong?