Personally, I don't knowingly open a trade before a major news event. If I'm already in one, I close it much earlier than 30 seconds before the news, and then decide afterwards whether or not to re-open.

I find "trading the news" particularly risky and unpleasant: it seems to me that there are so many spikes in both directions that it's pretty easy to have the overall direction right and still lose money by being stopped out.

It depends what type of trading you're doing, how major the news event is, and on various other variables?

MacroNinja 

It would seem as if your current trading strategy is potentially not building in any bias or calculation as to how to data from the news release will play out. If that's the case, then probably best to just close the position. However let's say for example that you knew for sure or a very high likelyhood that PMI data was going to come out at a 10% miss against consensus forecast. While everyone else might be surprised, you are not. If that's the case, why would you not hold the proper position ahead of time? Or let's say you knew that EPS was a huge beat at 1.50 versus a consensus forecast of .75. Again, if you've done the homework and you feel the analysis is correct, get paid for the work. If however, you're simply trading technicals and you have no edge when it comes to the fundamentals, then step aside and let the fundamental players make the valuation call, and after the dust settles and all the other technical traders have piled back into the market, then you get back to work.

You make a good point about incorporating some of the what-ifs related to data releases in the trading strategy - however I find myself in disagreement about knowing for sure or high likelihood that a certain data release is coming out one way or another. How could I ever forecast that with surety/high likelihood?!

A few things to keep in mind:

1) All banks have economists on staff that are culling through reams of data to come up with those forecasts. So when Bloomberg.com is showing a survey of those forecasts, it's Goldman's economist chiming in with a vote, Deutche's economist chiming in, etc. If you want to trade fundamentally, you have to start looking at the data as well. Or as a retail trader, simply stay out of the big boy's swimming pool.

2) The drawback (and therefore sometimes advantage) of these forecasts, is that as an economist, you want to stay somewhere in the consensus. If for example the range of apple EPS is 1.75-1.91, and you say that your forecast is 3.80 or .60, AND you're wrong, then you're going to get fired. So most of the time they try to stay within a consensus range. As long as your within consensus, you can still be wrong but not fired.

Additionally, these guys like to follow simply trends, which at some point don't make sense. e.g. NFP averages 200, 200, 200, so the forecast must be a straight line 200 right? Are there even 200k people to hire in a 4.9% unemployment market?

Bottom line, either get really good at the nitty gritty fundamental analysis, or just sit aside and watch the news as a spectators event. But of course, if you want to go pro, then this becomes unavoidable.

MacroNinja 

A few things to keep in mind: 1) All banks have economists on staff that are culling through reams of data to come up with those forecasts. So when Bloomberg.com is showing a survey of those forecasts, it's Goldman's economist chiming in with a vote, Deutche's economist chiming in, etc. If you want to trade fundamentally, you have to start looking at the data as well. Or as a retail trader, simply stay out of the big boy's swimming pool. 2) The drawback (and therefore sometimes advantage) of these forecasts, is that as an economist, you want to stay somewhere in the consensus. If for example the range of apple EPS is 1.75-1.91, and you say that your forecast is 3.80 or .60, AND you're wrong, then you're going to get fired. So most of the time they try to stay within a consensus range. As long as your within consensus, you can still be wrong but not fired. Additionally, these guys like to follow simply trends, which at some point don't make sense. e.g. NFP averages 200, 200, 200, so the forecast must be a straight line 200 right? Are there even 200k people to hire in a 4.9% unemployment market? Bottom line, either get really good at the nitty gritty fundamental analysis, or just sit aside and watch the news as a spectators event. But of course, if you want to go pro, then this becomes unavoidable.

Thanks, I don't disagree with the post above but, respectfully, that doesn't answer my earlier question.

Bottom line in the current context, close the trade because there is no basis or edge for the "instinct" that the price will move lower. If you can quantify the edge for price moving lower, the target if you are correct based upon the results of the data, and the risk exposure with the slippage, then by all means hold the trade.

If you were sitting on the pound desk, and I came up to you this morning to you why you are holding the trade going into CPI, and you're response was "because I think price will go down after the news", and then I ask you what is the consensus, CPI forecast, what is your CPI forecast, what data are you basis the differing forecasts, how will variance play into your targets, and how much of your down move has already been "baked" into the current price, how are you hedged if you're wrong, etc, and your answer as the trader is "I'm not sure"...then the risk manager probably comes by to turn off the computer and closes out the for you trade.

MacroNinja 

Bottom line in the current context, close the trade because there is no basis or edge for the "instinct" that the price will move lower. If you can quantify the edge for price moving lower, the target if you are correct based upon the results of the data, and the risk exposure with the slippage, then by all means hold the trade. If you were sitting on the pound desk, and I came up to you this morning to you why you are holding the trade going into CPI, and you're response was "because I think price will go down after the news", and then I ask you what is the consensus, CPI forecast, what is your CPI forecast, what data are you basis the differing forecasts, how will variance play into your targets, and how much of your down move has already been "baked" into the current price, how are you hedged if you're wrong, etc, and your answer as the trader is "I'm not sure"...then the risk manager probably comes by to turn off the computer and closes out the for you trade.

That is all good, well, understood and agreed. My original question still stands:

Quoting 

However let's say for example that you knew for sure or a very high likelyhood that PMI data was going to come out at a 10% miss against consensus forecast. While everyone else might be surprised, you are not.

How can you know for sure or with a very high likelihood that a certain data release is going to (say) be a 10% miss against consensus? Can you outline what the process would be in order to go about it?

There unfortunately is not a very short answer to that question. In general though, you have to take the data release you are trying to measure (for example CPI), build a model of the component parts to go into calculating the CPI, and use data inputs to that model to see how well you are able to simulate the government data, and then that will determine if your model has an accuracy in feeding you the correct number. Then anytime you get a different number from your model (which you have validated can correctly model the government stats), you start to lean on your forecasted number that you know is more accurate because the consensus forecasters simply have got it wrong.

Sometimes for example, this is a bit easier with something like earnings reports. e.g. if your job is to forecast # units of iPads and iPhones sold, and you can try to back into how many units the suppliers shipped for assembly, then you can start getting an idea of if the earnings forecast are even within a realm of possibility based upon what the orders are.

MacroNinja 

There unfortunately is not a very short answer to that question. In general though, you have to take the data release you are trying to measure (for example CPI), build a model of the component parts to go into calculating the CPI, and use data inputs to that model to see how well you are able to simulate the government data, and then that will determine if your model has an accuracy in feeding you the correct number. Then anytime you get a different number from your model (which you have validated can correctly model the government stats), you start to lean on your forecasted number that you know is more accurate because the consensus forecasters simply have got it wrong. Sometimes for example, this is a bit easier with something like earnings reports. e.g. if your job is to forecast # units of iPads and iPhones sold, and you can try to back into how many units the suppliers shipped for assembly, then you can start getting an idea of if the earnings forecast are even within a realm of possibility based upon what the orders are.

Thanks MacroNinja - I see the point you're trying to make - however I struggle to agree with it.

If being able to predict reasonably accurate results can be modeled, then why do we have the (sometimes widly) differing analyst views that give rise to the high/low estimates?

I attach an example of what I'm talking about. These were the NFP estimates for last November, ranging from a low of 75k to a high of 240k



This means that at least one of the analyst surveyed projected 75k and at least one projected 240k. Generally the analysts surveyed belong to well-known financial giants (UBS, Morgan Stanley, HSBC, Goldman, Deutsche, etc.) who undoubtedly would have the resources to model the estimates in the way you are suggesting.

Some of these analysts work on this kind of estimates day in day out I would bet - i.e. they don't trade like we do - so if we expected them to be able to issue accurate estimates it would be reasonable to assume their results to vary by little - yet we do have these gaps.

Why is that?

The question to answer your question would be whether or not an economist or analyst is giving a forecast to predict a specific model's outcome or if they are trying to describe what they think happened in the world. For example, for ADP's payroll report, are they trying to explain why they saw as the employment change or as they trying specifically to predict what the government's NFP will forecast?

Even different Fed districts has different models to forecast inflation and GDP. (The Atlanta Fed most notably uses a different model), and then on top of that the Fed itself for their formal decision making uses a different model than that technocrats at the US Government. So when the Atlanta Fed puts out thier GDP forecasts, are they saying that the other Fed districts and the BEA did the math wrong within their model, or is the Atlanta Fed possibly saying that they see and calculate the world a bit differently leading to a different number?

Regardless, I think we've veered quite a ways off your original, original question. All you have to do is trade profitably right? If you think they have an edge with the data, trade it. If not...maybe don't trade it -- it's whatever works for your risk model. Everything else is academic.

MacroNinja 

The question to answer your question would be whether or not an economist or analyst is giving a forecast to predict a specific model's outcome or if they are trying to describe what they think happened in the world. For example, for ADP's payroll report, are they trying to explain why they saw as the employment change or as they trying specifically to predict what the government's NFP will forecast? Even different Fed districts has different models to forecast inflation and GDP. (The Atlanta Fed most notably uses a different model), and then on top of that the Fed itself for their formal decision making uses a different model than that technocrats at the US Government. So when the Atlanta Fed puts out thier GDP forecasts, are they saying that the other Fed districts and the BEA did the math wrong within their model, or is the Atlanta Fed possibly saying that they see and calculate the world a bit differently leading to a different number? Regardless, I think we've veered quite a ways off your original, original question. All you have to do is trade profitably right? If you think they have an edge with the data, trade it. If not...maybe don't trade it -- it's whatever works for your risk model. Everything else is academic.

I agree, everything else is academic... the original message I got was 'if you do the homework you will know with high likelihood where a data release might be' (correct me if I got it wrong), which I'm still not convinced about...

I have absolutely no fear of working hard. There have been days where I worked 32 hours straight to achieve something I needed to get done. What I loathe is working hard for the sake of working hard, i.e. I'm not a fan of time-wasting.

I'd like to turn the whole topic on its head though. While I disagree that it is possible to predict a data release in advance, what I am doing is building a database that includes certain historic data releases and how the market reacted to them. This should give me, over the long term, a good idea as to what analysts predictions were, how the variance of the actuals was, and how price behaved accordingly. In a way that accounts for your initial suggestion, i.e. being able to factor in certain outcomes while in a trade.

Where this differs, though, is that rather than trying to predict what the data release will be, it should provide an understanding of how market participants interpreted and reacted to whatever comes out, whether in line or above/below expectations. I hope it makes sense.