What do you want to simulate using Monte Carlo? If you want to determine "ever" probabilities you can use Hoadley's: http://www.hoadley.net/options/barrierprobs.aspx? You can visualize and better understand options' sensitivities using this calculator: http://www.ftsweb.com/newfts/opsens.htm
I really want to see how fast (forgive me for this) and how furious volatility has to change to make the OTM options with a higher net vega more attractive than the ATM options.
You should give a real example of potential trade (price , strike(s) , vols and time to exp) , then someone can answer. The vega gain on far OTM might become only a paper profit , because of the bid/ask spread. If a 15 cents call[put] doubles because of vega , what is the actual profit on the 10 cents spread ? Plus the commissions diff on # of contracts(vs. ATM)
Wherever the pay-off is positive the 10 CUYQH (~ATM) perform better than the 26 CUYQG (OTM). The bottom graph, where the IV increased with 20%, shows that the pay-off is less which means that the OTM performs better. The two positions were sized for the same risk of ~$830. <img src=http://www.elitetrader.com/vb/attachment.php?s=&postid=1030697>
I think this is the heart of the argument combined with the much higher net Theta. This is a good explainer for the layman. It is just "faux vega."
a bit of subject... I don't use "increase of IV by x%" , going stickily by BP gains. If stock with IV of 30 increases 20% vega gains=600bp , but if stock with IV of 50 increases the same 20% , than vega gain is 100bp. Big difference in PnL. I do miss those days when stocks with IV of 100 use to gain 30% (300bp points) on pre-event vols spike...My profit per long combo is down by half since 2001.