Exness Spreads Hour by Hour — Server Time Read in Tbilisi — Georgia

Measured spreads — all 23 instruments in one table

Typical = the median of all readings. The min–p90 strip shows how far the spread stretched under load; a stable in sample badge means the spread never moved across the whole sample — on Exness Standard many instruments are quoted with stable target spreads, so identical min, median and p90 is expected there, not an error. Forex pairs are quoted in pips; metals, indices, energy and crypto in points — the cost column converts each instrument into the dollar cost of opening the selected lot size at the typical spread. Confidence dots: ●●● at least 2,000 readings, ●● at least 800, ● below that — treat single-dot rows as indicative. Dollar costs differ in magnitude because contract sizes differ: 1 lot of JP225 (Nikkei 225) is a contract of just 1 — a fraction of other instruments’ notional — so a cost of a few cents per lot is genuine, not an error; the “cost vs daily move” column is the fair cross-instrument comparison. Server Exness-MT5Trial11, captured 12 Aug 2026, 07:55 server time.

Spread through the day

Holding overnight — measured swaps

Values shown per 0.1 lot — the lot toggle above rescales this table too.

Read from the platform’s contract specifications at the same capture (2026-08-12). Negative = a nightly charge, positive = a nightly credit; on the triple-swap day three nights are applied at once. “Held 5 nights” counts 7 charges where the instrument has a triple-swap day — a Monday→Saturday hold crosses it once; energies (US Oil, UK Oil) have no triple day, so five nights mean five charges — the realistic cost of a week-long position, not the per-night teaser. See swap rates for how swaps work.

“From 0.0 pips” — which account is that, really?

The advertised “from 0.0” belongs to the Raw Spread account, which adds a fixed commission per side. This page measures the Standard account — the fair comparison is the all-in cost per trade:

Median spread, capture by capture (28 captures)

One capture per day from the same MT5 feed. A narrow range means the median spread held steady across the period. Spreads may fluctuate and widen due to volatility, news, and market conditions.

How this was measured

Which local hours this capture actually saw

The hour chart is sampled rather than modelled: bars cover the stretch the capture window reached and hatched columns are the hours it did not. In server terms that window ran from 07:00 to 20:00. Converted to the Tbilisi clock it runs from late morning to midnight, which happens to be the part of the day most people here are awake for.

The hours without a bar are therefore the local small hours, roughly 01:00 to 10:00. They are not zero spreads and not a hole in the market; they simply fall outside this capture. Reading the chart with the local labels in mind stops a hatched column from being mistaken for a quiet market.

That also sets the honest limit of the page for anyone here who trades early. The measured picture is at its strongest for the Tbilisi afternoon and evening and says nothing at all about the local dawn.

Midnight, and the hour after it

The widest single reading in this sample sits at 20:00 server, which is exactly midnight on the Tbilisi clock. The band the chart marks as thin around the daily rollover starts an hour later, at 01:00 local, and runs to 04:00, where the server day turns over and the new daily candle opens.

That hour is also where the nightly holding charge is applied. A position still open when the local date has turned has crossed the boundary; one closed at eleven in the evening has not. The swap table further down counts nights, and this is the moment at which it counts them.

Anyone judging entry cost from the table should notice which local hour a habit falls into. A routine of opening positions just before bed in Tbilisi is a routine of opening them in the widest hour this capture recorded.

The calendar behind the swap columns

The triple-day column is a calendar fact rather than a price one. The measured rows show currency pairs applying three nights on Wednesday and index instruments doing it on Friday, so an identical overnight hold costs one night or three depending only on which day it crosses.

The held-five-nights column follows the same arithmetic. A position opened on a Tbilisi Monday afternoon and left until Saturday crosses the triple day once, which is why five nights of calendar produce seven charges. Energy rows carry no triple day and charge once per night.

The weekend is the other calendar edge. Quotes stop on Saturday morning local time and resume in the small hours of Monday, so a Friday evening position waits through two days during which the table has nothing new to report and the price can still be different when it does.

Summer time moves the bands, not the offset

The session bands under the hour chart are anchored to summer settings in London and New York. Georgia does not change its clocks, so when those places move theirs the bands slide against the local hour while the server offset stays at four in both directions.

In practice one line of a Tbilisi trading schedule is seasonal and the rest is not. The conversion from server to local is permanent; the local hour at which the overlap begins is the part that has to be redrawn in spring and autumn.

Convert the hours before comparing the numbers

1. Add four hours to every server hour printed on this page to reach the Tbilisi clock.
2. Mark the sampled window: 07:00 to 20:00 server is 11:00 to midnight local.
3. Read hatched columns as unmeasured local small hours rather than as quiet trading.
4. Note that the widest reading of the sample sits at local midnight and that the thin band around the rollover runs from 01:00 to 04:00 in Tbilisi.
5. Check the triple-day column against the weekday a position will actually cross, not against the day it was opened.
6. Plan around the weekend pause: from Saturday morning to the small hours of Monday the table has nothing new to say.

Hours here are the platform server clock from the same capture that produced the spread figures on this page; the Georgian offset is four hours and it does not shift with the seasons.

The session bands on the Tbilisi clock

Band edges are the indicative summer-time anchors printed with the chart; the local column follows a fixed UTC+4 offset. This capture sampled 07:00 to 20:00 server, so the Asia band and the thin band were outside the window.

Which line here answers the question what if I did otherwise

Three lines of this table are decisions and three are not. Only the first three are worth arguing with.

Related Exness pages