Economically, a forward and a future on the same underlying and maturity are priced by the identical cost-of-carry argument. Operationally, they are built completely differently, and the difference is entirely about when the gain or loss is realized. A forward is an OTC, bilateral contract that settles once, in full, at maturity, every dollar of gain or loss accrues silently until that one date, which is exactly why Worked Example 2 needed a separate "value this contract mid-life" step: before maturity, an outstanding forward has a replacement-cost value, not a cash value, computed as the present value of the difference between today's fair forward price for the remaining time and the rate you originally locked in:
V_long = Notional × (F_today,remaining − F_contracted) × DF(remaining time)
How do mark-to-market and margin calls work?
A future, by contrast, is an exchange-traded, centrally-cleared contract that is marked to market daily: at the end of every trading session, the exchange compares that day's settlement price to the previous one and moves cash (called variation margin) directly into or out of every position's margin account, for every contract, every day. Worked Example 1 walks a small Nifty futures position through three days of settlement prices and shows the defining property of daily settlement: the sum of the daily cash flows always equals the total price change from entry to exit, no matter how the path wiggled in between, mark-to-market redistributes when you receive the P&L, never how much, in total, you receive.
This daily cash-flow mechanism is why futures carry an initial margin (a good-faith deposit posted at trade inception) and a maintenance margin (a lower threshold (commonly set as a fraction of initial margin) below which the exchange issues a margin call, requiring the account to be topped back up to the initial margin level, not merely back to maintenance). Practice P13 walks a full multi-day margin sequence including a maintenance breach, precisely because this is the point most first-time learners get backwards: a call takes you back to the top, not to the floor that triggered it. This daily-settlement design is also what makes exchange-traded futures functionally free of the counterparty credit risk an uncleared OTC forward carries, a genuinely valuable feature, at the cost of daily liquidity/cash-management pressure a forward does not impose.
Advertisement
Why should you recompute derivatives prices yourself?
Of every region in this program, derivatives pricing and valuation is arguably the one carrying the corpus's highest AI-over-trust risk, and it is worth saying plainly, once, exactly here, because mark-to-market is where the danger is most acute. A wrong discount curve, a swapped sign on which party is long, or a day-count fraction silently using 360 where the desk quotes on 365 all produce an output that looks completely identical to a correct one: a number, to two decimal places, delivered with total fluency. There is no "this looks suspicious" tell a paragraph of prose can carry, because the paragraph is not where the error lives, the arithmetic is, and the arithmetic is invisible in the answer's tone (exactly AI0.01's point about a model's confidence carrying no information about its correctness). That is precisely why every ai-augment-json callout in this module routes its verify step through rebuilding the discount factors or the settlement formula yourself and checking it against a hand or independently-coded computation, never through asking the model whether it is sure, and never through accepting a swap valuation, a curve, or an FRA rate that you have not personally reconciled against the shortcut/consistency check that section supplies.