Apologies for posting this query here but I have recieved little response in the General Forum.I hope some of you market professionals can help answer a few queries to do with my student research project on "swap pricing".I have collected some high frequency indicative quote data via Reuters on (i) sterling deposits (1 day, 7 day, 1 month, 2 month and 3 month), (ii) short-sterling futures on quarterly expiration cycle, and (iii) one-year sterling swap (in fact the zero coupon swap where the fixed rate is paid/received at the end of the swap but the floating ratea are received/paid at quarterly intervals). I use this data to examine the pricing of the swap at one-minute intervals over a 12-month period.I calculate the theoretical swap rate using the FRN method. This calculation is done at one-minute intervals using a cash/futures strip calculated using the most most recent midquotes on deposits and futures (cash-to-first-futures then successive futures rates).Not suprisingly, I find that the discrepancy between the actual swap midquote and the theoretical swap midquote is very small, mean of 0.3 bp, standard deviation of 1.5 bp. The 'mispricing' exceeds +/- 3 bp only 1% of the time. However I have a huge number of observations (over 150,000 during the hours of trading Euronext.LIFFE over a year) and sometimes the mispricing can reach +/- 15 basis points. The majority of these do not disappear when I exclude observations where the swap or deposit quote is more than 15 seconds old. I attribute the presence of this 'mispricing' to the noise inherent in a times series of Reuters indicative quotes from an OTC market.I have not made convexity adjustments to the futures rates prior to calculating the theoretical swap rate since (i) I don't have any data on the size of the adjustment, and (ii) some sources say that this is "minor" for futures contracts with less than 12 months to expiration.Question 1:What is the typical convexity adjustment for (a) short-sterling futures contracts with (i) 3-months, (ii) 6-months and (iii) 9-months to expiration., and (b) the one-year zero coupon swap. I realise that these adjustments will change over time but some estimates would be appreciated.Question 2:From a purely theoretical perspective, how would a trader arbitrage genuine 'mispricing' of the zero coupon swap? The mismatch in interest payment frequencies seems to rule out using futures contracts in the arbitrage. Since a zero coupon swap represents offsetting positions in one-year fixed rate and floating rate bonds, buying (selling) a swap and selling (buying) a synthetic swap via the bond market would appear to be the appropriate strategy when mispricing appears. Is this the right approach? If so, what transactions costs (in terms of bps in the bond and swap markets) would be involved? Also what other risks/costs are involved? If this type of arbitarge strategy is not the right one, what would be more a more appropriate strategy? A simple example would help!

From a purely theoretical perspective, how would a trader arbitrage genuine 'mispricing' of the zero coupon swap?In practice, a trader will simply execute the trade using swap/futures strip when an 'arb' is identified, and wait until the market realigns itself and the arb dissapears (as you said, it usually does). At this point the trader will have the position unwound at a profit.If the 'arb' positions widen even further, the trader takes a mark-to-market loss for the time being and waits longer, and if that loss is too big, the position may need to be closed out at a loss.One can treat your arb-analysis as a trading signal.

Thanks "inarrears".I understand what you say but how do you execute a "perfect" arbitrage when a zero coupon swap is involved? I can grasp how an arb could use futures to arbitrage mispricing in a conventional swap (i.e. one involving quarterly exchanges of fixed/floating rate cashflows), but can a zero coupon swap be arbitraged in the same way?

what does it matter whether you trade an IMM futures strip vs a 1Y swap that pays quarterly money, semi-bond, or annual money? As long as you know that the 'arbitrage mispricing' will kick in (as you have observed in your study), and the market will get back in line, who cares whether you are executing a 'perfect' or approximate 'arbitrage'? You can just unwind your position once the market is back in line and profit. there is really a fine line between identifying an 'arb' based on yield curve modelling and a relative value trading signal.no one cares about executing a 'perfect' arb. why bother? you can't hedge everything perfectly, nor anyone tries to, when the typical trader runs a big book of a thousands of swaps with mismatched fixing dates. taking on basis risk is just a part of the business.

Thanks again "inarrears" for those insights in arbitrage.Can you provide approximate "convexity" estimates (Q1)?