Frequency describes a process, not quality
Distinguish trade count from turnover. One hundred small operations may have less monetary turnover than one large portfolio restructuring, and one decision may execute through several trades. Cost analysis needs bought and sold volume, tariff basis and achievable price, not only rows in a broker export.
A fictional direct-cost model
Let capital be 100,000 roubles and buying then selling a full equivalent cost 0.15%, excluding taxes. Twelve cycles at a constant base cost 1,800 roubles per year; 120 cost 18,000. This is a linear control model: capital is unchanged, costs are not capitalized and every cycle is identical. It is not a current intermediary tariff.
For each case specify commission, spread, slippage, data fees and possible financing. Fixed costs divide differently from proportional costs. A “free” platform condition tied to activity belongs to the full economics and does not disappear from it.
The spread is not always the visible number
Best displayed bid and offer prices apply to limited quantity. A large order may fill across levels, and price may move between submission and execution. A last-trade calculation therefore does not guarantee attainability. For a short horizon, even a small fraction of a percent can be comparable to the gross effect studied. Less frequent trading may reduce turnover influence but introduces price movement between observations, duration of risk, distributions and changed hypotheses.
Infrastructure consumes resources too
High decision frequency may require precise timing, resilient connectivity, event sequencing, outage monitoring and journal recovery. Research-export latency and real-system latency can change the achievable result. This is about constraints, not a high-frequency trading construction manual.
Include time spent. Hours of monitoring are not broker commission, but they are part of the economic cost for a person. If comparing alternatives, state whether and how time is valued; unpaid labor is not an infinite free resource.
What historical research supports
Barber and Odean studied private-investor accounts at a US discount broker in the 1990s. Active trading was associated with a substantial loss after costs, and the most active group lagged. This concerns a defined sample and does not prove every frequent trade is pointless; it shows why gross results and participant confidence are insufficient.[1]
Use new data and current tariffs for a new situation. Do not rewrite historical commissions as 2026 conditions or treat a group average as a personal forecast. Planned rebalancing, a payment need and trying to guess the next tick solve different problems.
A checkable final table
For each approach, bridge gross result to net: price and distribution income, all costs, then taxes and limits on applying their calculation. Show worst period, turnover and when cash was needed. Do not hide missing data behind a final percentage. A justified process with a visible full price beats a faster or slower label.
Same capital, different processes
Compare two fictional activities on 1 million monetary units, without leverage or taxes. A three-year business holding produces 80,000 gross in one control year, with two full turnover cycles and 0.10% cost per cycle: 78,000 after direct costs; research takes 40 hours. A day-horizon execution rule also produces 80,000 gross, but with 100 cycles at 0.10%: costs are 100,000 and net result is −20,000; observation takes 600 hours.
At a notional 500 units per hour, time costs are 20,000 and 300,000, leaving economic residuals of 58,000 and −320,000. This is a chosen alternative, not an account debit. The examples do not prove investing always wins or that a trading rule cannot have a larger gross effect.
A calm market: when doing nothing is the rule
Suppose a plan targets a 50% share of a risky asset, with a 45–55% review band, no new money and no payments. The current share is 51% and there are no new verified facts. Boredom or fear of missing out is not a new basis: the pre-defined action is to make no unscheduled trade and record the absence of a trigger. This does not claim that tomorrow’s price will remain unchanged.
A planned contribution or a share outside the announced band is a counterexample: it can justify a transaction after calculating the new target, deviation and costs. Calm prices are neither a buy signal nor a universal reason to hold cash.
HFT as a system and the emergency stop
A high-frequency process is an automated loop: receive event → restore market state → check signal and limits → send message → confirm or reject → reconcile position. Inputs need sequence, timestamps and gap detection; snapshot plus update stream must reconstruct one state. Limits, rejection handling, identifiers and execution reconciliation follow. A cancellation request is not proof that an order is canceled.
If a signal lives for 500 microseconds while the 99th-percentile path to confirmation is 900 and the mean is 200, the mean does not prove every signal is usable. These figures are fictional.
A hypothetical system permits messages for 2 million units per second; failure detection and actual stop take 0.5 seconds, during which an additional 1 million may be sent. This is nominal message volume, not proven loss or a position limit. Restart requires journal, execution and residual-order reconciliation.
A 2020 Nasdaq FAQ describes a kill switch as a post-trade best-efforts process that disables ports and cancels covered orders after a limit, with exceptions.[1](#source-cvclose-nasdaq-kill) It is a historical technical example, not a 2026 interface or setup instruction. Economic viability is separate: 300,000 fixed annual cost divided by 0.20 per completed cycle requires 1.5 million cycles; at 0.10 it requires 3 million.
Sources and boundaries
[1] [Trading Is Hazardous to Your Wealth](https://faculty.haas.berkeley.edu/odean/papers%20current%20versions/individual_investor_performance_final.pdf): historical clients of one US broker, not a 2026 tariff or forecast. The material does not promise HFT success, assess current tariffs or describe admission to live trading.