NISM Professor

Spot price polling

Also written Polling · Spot polling · Polled spot price

The statistical process by which an exchange collects physical-market prices from an empanelled panel of traders and users, strips out outliers, and publishes a single daily spot price.

In plain language

A commodity futures contract needs a spot price to anchor to. For equities that is trivial — the cash market sits on the same exchange. For commodities it is not, because the cash market is a few hundred mandis scattered across the country, none of them reporting to anybody.

So the exchange asks. It maintains a panel of traders and users of each commodity, collects their quotes, and reduces the answers to one number.

That number matters more than it looks. It tells traders how much contango or backwardation is built into the futures price, it feeds the Final Settlement Price of every physically settled contract, and for Options on Goods it is the settlement price on which exercise turns.

How it works

Polling is defined as compiling price information from a cross-section of market players from different centres across the country. Prices are collected from empanelled polling participants — traders and users of the particular commodity — and exchanges review the panel regularly, adding active participants and removing inactive ones.

What gets quoted is specific: ex-warehouse prices, inclusive of Mandi Cess but exclusive of GST. On NCDEX, participant-wise prices for each commodity are disclosed twice daily, at around 1:30 pm and again at around 4:00 to 5:00 pm.

The reduction to one number runs in two stages:

  1. Bootstrapping removes extreme quotes. A median is arrived at and outlier prices are dropped from further processing.
  2. An algorithm then runs across the surviving data on the principle of least error — closely related to the least squares method of regression. Random candidate prices are compared on the standard deviation of the polled set around them; the candidate with the higher deviation is discarded and another tried, repeatedly, until the price with the least standard deviation against the polled data survives. That is the polled spot price for that time on that day.

Where the polled quotes are tightly ranged, the result comes out close to the simple average, median or mode. It is only in a scattered market that the method earns its keep.

Exchanges must disclose, for each commodity: contract details, the polling mechanism, how prices are arrived at, whether taxes and levies are included, whether polling has been outsourced and to whom, the criteria for selecting participants, and anything else that improves transparency.

A worked example

RM seed polled at the 1:30 pm run, in rupees per quintal, from seven empanelled participants:

ParticipantQuote
A5,480
B5,495
C5,500
D5,505
E5,515
F6,240
G4,900

The simple average of all seven is Rs 5,376 — dragged down by a single stale quote at 4,900 and not rescued by the 6,240. It describes nobody's market.

Bootstrapping removes F and G as outliers around the median of 5,500. The algorithm then searches the surviving five quotes for the price with the least standard deviation against them, and lands on roughly Rs 5,500.

On a 10 MT = 100 quintal lot, the difference between the naive average and the polled price is:

(5,500 - 5,376) x 100 quintals = Rs 12,400 per lot

Carry that into settlement across three days and it moves the Due Date Rate, every delivery invoice struck on it, the exercise decision on every Option on Goods strike near 5,500, and the replacement cost in any delivery default. Rs 12,400 a lot is what the statistics are for.

Why NISM asks about it

Chapter 3 (Commodity Futures), section 3.10, with the Options on Goods dependency in Chapter 4 section 4.6 and the disclosure obligations in Chapter 6 section 6.9. Expect questions on what polled prices include and exclude (Mandi Cess in, GST out), on bootstrapping and the principle of least error, and on the FSP being the simple average of the last three days of polled prices.

Common exam traps

  • Polled prices include Mandi Cess but exclude GST. Both halves get asked.
  • They are ex-warehouse prices, so freight to the buyer's location is not in them.
  • Bootstrapping is the outlier-removal step, not the final averaging step. The final number comes from the least-error search that follows it.
  • The polled price is not necessarily the median or the mean — it only coincides with them when the quotes are tightly clustered.
  • Polling is what makes Options on Goods possible at all, and also what forces the seven-strike CTM band, because a sampled price carries sampling error.
  • The panel is reviewed and changed over time. Polling participants are not a fixed list, and the exchange must publish its selection criteria.

Where this is taught

Free preparation for NISM Series XVI

Related terms

← All terms
Something look wrong? Report it