If you have X-Y data, this should be easy. If not, just from the image, in the top half, it appears that the portion of the curve at higher delT is heating and at lower delT is cooling. The curve is cut-off, so I can not tell for sure. You should look for structural relaxation of the alloy, or other endothermic processes on your curve, as they will appear on heating.
At maximum temperature, break curves into two parts. Then you need to identify peaks. Some basics you can see in https://www.researchgate.net/publication/257713217_Study_of_Carbide_Evolution_During_Thermo-Mechanical_Processing_of_AISI_D2_Tool_Steel.
Article Study of Carbide Evolution During Thermo-Mechanical Processi...
Hi I would hazard a guess that the top three curves are your heating curves. Check your run parameters as the top three curves start at RT and the bottom three end at 200 dC that should help you discriminate between them alternatively look at the three curves with the x-axis in time and check the temp at the beginning of your run. If the software allows you to view the steps individually, rather than the whole curve, you can then look at the heating and cooling steps separately with x-axis in temp to do your analysis and then export the steps as an ASCII and plot them in excel or origin etc. I would not analyse the peaks you are showing at > 1380 dC as I would suspect that these may be due to the run conditions and not sample changes - if you think they are sample changes re-run to + 50 dC from the end of the transition to allow the DTA signal to return to baseline
Why do you stop your measurement at ~1425 C if thats close to your transition? I recommend you make a longer measurement, at least to 1450 ˚C. Otherwise you can't claim there was a transition.