BDI & SCFI prediction (‘22/23W)

In the 23rd week, BDI and SCFI predicted values and actual values are compared and tested.

 

1.     Last Week’s forecasts

1-1.        BDI, Actual Vs. Predicted

Date	Last week Forecast	Actual BDI	MAPE(%)	Accuracy(%)
‘22/22W(5/30~6.3)	2,937.708	2,590.00	13.4%	86.6%

 

1-2.        SCFI, Actual Vs. Predicted

Date	Last week Forecast	Actual SCFI	MAPE(%)	Accuracy(%)
‘22/22W(5/30~6.3)	4,177.124	4,208.01	0.7%	99.3%

 The prediction accuracy of BDI and SCFI was 86.6% and 99.3%, respectively. The predicted value (2,937.708) and actual value (2,590.00) of BDI, and predicted value (4,177.124) and actual value (4,208.01) of SCFI all achieved high accuracy and trend. In the case of BDI, the rate of error in the forecast also increased to 13.4% because the freight rate fell more sharply than expected. Moreover, since the BDI rebounded slightly in the second half of the week, it is difficult to diagnose next week's (23 weeks) freight rate forecast.

  The average BDI MAPE value for the last 10 predictions is 4.8% and the prediction rate is 95.2%, and the SCFI average MAPE value is 0.7% and the prediction rate is 99.3%.

---

2.     BDI forecast

2-1. ARIMA, Time series analysis

 For the BDI forecast, I used weekly data (435 pieces) from January 3, 2014 to Jun 03, 2022 to predict the freight for three weeks in the future.

Date	Forecast	Trend(WoW)
‘22/23W(6/6~10)	2,241.961	↓
‘22/24W(6/13~17)	2,102.699	↓
‘22/25W(6/20~24)	2,106.618	↗

 As shown in the table and figure above, BDI (as of June 3rd: 2,590.00) plummeted by -15.7% compared to last week (as of May 27th: 3,072.60). In particular, on a daily basis, although the Cape market conditions recovered somewhat in the middle of the week, both Panamax and Supramax linear volumes decreased, negatively affecting the market conditions.

 As a result of the ARIMA model, it was predicted that the BDI would slightly rebound in the third week (25 weeks) after a decline for two consecutive weeks.

 

Ref) BDI & SCFI prediction (‘22/22W) (tistory.com)

BDI & SCFI prediction (‘22/22W)

---

2-2. Logistic regression analysis

 Logistic regression analysis was performed with the same data predicted by the ARIMA model previously.

 Next week's expected trend is 0.264, well below the baseline of 0.5. Since the BDI is on a downward trend after a sharp rise, econometrics also predicts a downward trend.

---

3.     SCFI Outlook

3-1. ARIMA, Time series analysis

 For the SCFI forecast, I used weekly data (633 pieces) from October 16, 2009 to Jun 03, 2022 to predict the future freight for three weeks.

Date	Forecast	Trend(WoW)
‘22/23W(6/6~10)	4,210.933	↗
‘22/24W(6/13~17)	4,213.438	↘
‘22/25W(6/20~24)	4,209.118	↘

 consecutive weeks following last week (as of May 27th: 4,175.35). The rate of change of SCFI increased from 0.3% two weeks ago to 0.78%, slowly dissolving market participants' doubts about the trend reversal of SCFI. If the uptrend continues with the increase in the uptrend next week, it can be said that a clear trend reversal has occurred.

---

3-2. Logistic regression analysis

 The predicted value of SCFI's logistic regression analysis model was 0.394, which predicted the next week's SCFI's freight direction to be down. However, in light of the fact that it is close to the baseline (0.5) and rising from the predicted value (0.393) two weeks ago, it can be seen that the time of trend reversal is slowly approaching.

---

3-3. SVR model

 

 The SVM model predicted a continued decline in SCFI rates next week. However, the model shows that the upward trend of freight rates will turn around after 24 weeks. All statistical models are showing that a fare conversion signal will happen soon.

Date	Forecast	Trend(WoW)
‘22/23W(6/6~10)	4,146.348	↘
‘22/24W(6/13~17)	4,164.223	↗
‘22/25W(6/20~24)	4,181.740	↗

Thanks.